Deep Learning Projects/natural-language-processing-with-disaster-tweets.ipynb

{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.10.12","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"import pandas as pd\ntrain_df = pd.read_csv(\"/kaggle/input/nlp-getting-started/train.csv\")\ntest_df = pd.read_csv(\"/kaggle/input/nlp-getting-started/test.csv\")","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:16.494008Z","iopub.execute_input":"2023-10-04T08:15:16.494367Z","iopub.status.idle":"2023-10-04T08:15:16.536579Z","shell.execute_reply.started":"2023-10-04T08:15:16.494341Z","shell.execute_reply":"2023-10-04T08:15:16.535149Z"},"trusted":true},"execution_count":114,"outputs":[]},{"cell_type":"code","source":"train_df.head()","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:16.538354Z","iopub.execute_input":"2023-10-04T08:15:16.538871Z","iopub.status.idle":"2023-10-04T08:15:16.552041Z","shell.execute_reply.started":"2023-10-04T08:15:16.538835Z","shell.execute_reply":"2023-10-04T08:15:16.550955Z"},"trusted":true},"execution_count":115,"outputs":[{"execution_count":115,"output_type":"execute_result","data":{"text/plain":"   id keyword location                                               text  \\\n0   1     NaN      NaN  Our Deeds are the Reason of this #earthquake M...   \n1   4     NaN      NaN             Forest fire near La Ronge Sask. Canada   \n2   5     NaN      NaN  All residents asked to 'shelter in place' are ...   \n3   6     NaN      NaN  13,000 people receive #wildfires evacuation or...   \n4   7     NaN      NaN  Just got sent this photo from Ruby #Alaska as ...   \n\n   target  \n0       1  \n1       1  \n2       1  \n3       1  \n4       1  ","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>id</th>\n      <th>keyword</th>\n      <th>location</th>\n      <th>text</th>\n      <th>target</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>1</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>Our Deeds are the Reason of this #earthquake M...</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>4</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>Forest fire near La Ronge Sask. Canada</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>5</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>All residents asked to 'shelter in place' are ...</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>6</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>13,000 people receive #wildfires evacuation or...</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>7</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>Just got sent this photo from Ruby #Alaska as ...</td>\n      <td>1</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"test_df.head()","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:16.553570Z","iopub.execute_input":"2023-10-04T08:15:16.553995Z","iopub.status.idle":"2023-10-04T08:15:16.572445Z","shell.execute_reply.started":"2023-10-04T08:15:16.553964Z","shell.execute_reply":"2023-10-04T08:15:16.571204Z"},"trusted":true},"execution_count":116,"outputs":[{"execution_count":116,"output_type":"execute_result","data":{"text/plain":"   id keyword location                                               text\n0   0     NaN      NaN                 Just happened a terrible car crash\n1   2     NaN      NaN  Heard about #earthquake is different cities, s...\n2   3     NaN      NaN  there is a forest fire at spot pond, geese are...\n3   9     NaN      NaN           Apocalypse lighting. #Spokane #wildfires\n4  11     NaN      NaN      Typhoon Soudelor kills 28 in China and Taiwan","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>id</th>\n      <th>keyword</th>\n      <th>location</th>\n      <th>text</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>0</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>Just happened a terrible car crash</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>2</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>Heard about #earthquake is different cities, s...</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>3</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>there is a forest fire at spot pond, geese are...</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>9</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>Apocalypse lighting. #Spokane #wildfires</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>11</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>Typhoon Soudelor kills 28 in China and Taiwan</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"train_df.target.value_counts()","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:16.573906Z","iopub.execute_input":"2023-10-04T08:15:16.575026Z","iopub.status.idle":"2023-10-04T08:15:16.590210Z","shell.execute_reply.started":"2023-10-04T08:15:16.574983Z","shell.execute_reply":"2023-10-04T08:15:16.589122Z"},"trusted":true},"execution_count":117,"outputs":[{"execution_count":117,"output_type":"execute_result","data":{"text/plain":"target\n0    4342\n1    3271\nName: count, dtype: int64"},"metadata":{}}]},{"cell_type":"code","source":"len(train_df) + len(test_df)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:16.593249Z","iopub.execute_input":"2023-10-04T08:15:16.594163Z","iopub.status.idle":"2023-10-04T08:15:16.609295Z","shell.execute_reply.started":"2023-10-04T08:15:16.594120Z","shell.execute_reply":"2023-10-04T08:15:16.607476Z"},"trusted":true},"execution_count":118,"outputs":[{"execution_count":118,"output_type":"execute_result","data":{"text/plain":"10876"},"metadata":{}}]},{"cell_type":"code","source":"# shuffle the training data \ntrain_data = train_df.sample(frac =1, random_state = 42)\ntrain_data.head()","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:16.610498Z","iopub.execute_input":"2023-10-04T08:15:16.611507Z","iopub.status.idle":"2023-10-04T08:15:16.628647Z","shell.execute_reply.started":"2023-10-04T08:15:16.611460Z","shell.execute_reply":"2023-10-04T08:15:16.627492Z"},"trusted":true},"execution_count":119,"outputs":[{"execution_count":119,"output_type":"execute_result","data":{"text/plain":"        id      keyword               location  \\\n2644  3796  destruction                    NaN   \n2227  3185       deluge                    NaN   \n5448  7769       police                     UK   \n132    191   aftershock                    NaN   \n6845  9810       trauma  Montgomery County, MD   \n\n                                                   text  target  \n2644  So you have a new weapon that can cause un-ima...       1  \n2227  The f$&amp;@ing things I do for #GISHWHES Just...       0  \n5448  DT @georgegalloway: RT @Galloway4Mayor: ÛÏThe...       1  \n132   Aftershock back to school kick off was great. ...       0  \n6845  in response to trauma Children of Addicts deve...       0  ","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>id</th>\n      <th>keyword</th>\n      <th>location</th>\n      <th>text</th>\n      <th>target</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>2644</th>\n      <td>3796</td>\n      <td>destruction</td>\n      <td>NaN</td>\n      <td>So you have a new weapon that can cause un-ima...</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>2227</th>\n      <td>3185</td>\n      <td>deluge</td>\n      <td>NaN</td>\n      <td>The f$&amp;amp;@ing things I do for #GISHWHES Just...</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>5448</th>\n      <td>7769</td>\n      <td>police</td>\n      <td>UK</td>\n      <td>DT @georgegalloway: RT @Galloway4Mayor: ÛÏThe...</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>132</th>\n      <td>191</td>\n      <td>aftershock</td>\n      <td>NaN</td>\n      <td>Aftershock back to school kick off was great. ...</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>6845</th>\n      <td>9810</td>\n      <td>trauma</td>\n      <td>Montgomery County, MD</td>\n      <td>in response to trauma Children of Addicts deve...</td>\n      <td>0</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"# visulize the data\nimport random\nrandom_index = random.randint(0, len(train_df) - 5)\n\nfor row in train_data[[\"text\", \"target\"]][random_index:random_index + 5].itertuples():\n    _,text, target = row\n    print(f\"Target: {target}\", \"(real disater)\" if target > 0 else \"(not real disater)\")\n    print(f\"Text: \\n {text} \\n\")\n    print(\"...\\n\")\n    ","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:16.629937Z","iopub.execute_input":"2023-10-04T08:15:16.631200Z","iopub.status.idle":"2023-10-04T08:15:16.639074Z","shell.execute_reply.started":"2023-10-04T08:15:16.631167Z","shell.execute_reply":"2023-10-04T08:15:16.638318Z"},"trusted":true},"execution_count":120,"outputs":[{"name":"stdout","text":"Target: 0 (not real disater)\nText: \n @whvholst @leashless And this is a structural problem rather than just a failure of competence by traditional soc democratic parties. \n\n...\n\nTarget: 0 (not real disater)\nText: \n I SCREAMED 'WHATS A CHONCe' http://t.co/GXYivsWki7 \n\n...\n\nTarget: 0 (not real disater)\nText: \n @FaZe_Rain all hail the Cloud \n\n...\n\nTarget: 1 (real disater)\nText: \n Typhoon Soudelor taking dead aim at Taiwan http://t.co/3Ac5wuy1R0 \n\n...\n\nTarget: 0 (not real disater)\nText: \n Violators of the new improved Reddit will be shot into the sun. Survivors will be hurled into a black hole and then nuked. \n\n...\n\n","output_type":"stream"}]},{"cell_type":"code","source":"# split the data inot training and validation set\nfrom sklearn.model_selection import train_test_split\n\ntrain_sentences, val_sentences, train_labels, val_labels = train_test_split(train_data[\"text\"].to_numpy(),\n                                                                           train_data[\"target\"].to_numpy(),\n                                                                           test_size =0.1,\n                                                                           random_state = 42)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:16.640407Z","iopub.execute_input":"2023-10-04T08:15:16.641361Z","iopub.status.idle":"2023-10-04T08:15:16.653370Z","shell.execute_reply.started":"2023-10-04T08:15:16.641332Z","shell.execute_reply":"2023-10-04T08:15:16.652414Z"},"trusted":true},"execution_count":121,"outputs":[]},{"cell_type":"code","source":"len(train_sentences), len(val_sentences), len(train_labels), len(val_labels)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:16.654554Z","iopub.execute_input":"2023-10-04T08:15:16.655452Z","iopub.status.idle":"2023-10-04T08:15:16.670202Z","shell.execute_reply.started":"2023-10-04T08:15:16.655424Z","shell.execute_reply":"2023-10-04T08:15:16.669360Z"},"trusted":true},"execution_count":122,"outputs":[{"execution_count":122,"output_type":"execute_result","data":{"text/plain":"(6851, 762, 6851, 762)"},"metadata":{}}]},{"cell_type":"code","source":"train_sentences[:10]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:16.671516Z","iopub.execute_input":"2023-10-04T08:15:16.672180Z","iopub.status.idle":"2023-10-04T08:15:16.686976Z","shell.execute_reply.started":"2023-10-04T08:15:16.672142Z","shell.execute_reply":"2023-10-04T08:15:16.685738Z"},"trusted":true},"execution_count":123,"outputs":[{"execution_count":123,"output_type":"execute_result","data":{"text/plain":"array(['@mogacola @zamtriossu i screamed after hitting tweet',\n       'Imagine getting flattened by Kurt Zouma',\n       '@Gurmeetramrahim #MSGDoing111WelfareWorks Green S welfare force ke appx 65000 members har time disaster victim ki help ke liye tyar hai....',\n       \"@shakjn @C7 @Magnums im shaking in fear he's gonna hack the planet\",\n       'Somehow find you and I collide http://t.co/Ee8RpOahPk',\n       '@EvaHanderek @MarleyKnysh great times until the bus driver held us hostage in the mall parking lot lmfao',\n       'destroy the free fandom honestly',\n       'Weapons stolen from National Guard Armory in New Albany still missing #Gunsense http://t.co/lKNU8902JE',\n       '@wfaaweather Pete when will the heat wave pass? Is it really going to be mid month? Frisco Boy Scouts have a canoe trip in Okla.',\n       'Patient-reported outcomes in long-term survivors of metastatic colorectal cancer - British Journal of Surgery http://t.co/5Yl4DC1Tqt'],\n      dtype=object)"},"metadata":{}}]},{"cell_type":"code","source":"train_labels[:10]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:16.688357Z","iopub.execute_input":"2023-10-04T08:15:16.688761Z","iopub.status.idle":"2023-10-04T08:15:16.699597Z","shell.execute_reply.started":"2023-10-04T08:15:16.688723Z","shell.execute_reply":"2023-10-04T08:15:16.698958Z"},"trusted":true},"execution_count":124,"outputs":[{"execution_count":124,"output_type":"execute_result","data":{"text/plain":"array([0, 0, 1, 0, 0, 1, 1, 0, 1, 1])"},"metadata":{}}]},{"cell_type":"markdown","source":"## Text Vectorization","metadata":{}},{"cell_type":"code","source":"import tensorflow as tf\nfrom tensorflow.keras.layers import TextVectorization","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:16.700763Z","iopub.execute_input":"2023-10-04T08:15:16.701079Z","iopub.status.idle":"2023-10-04T08:15:16.711043Z","shell.execute_reply.started":"2023-10-04T08:15:16.701053Z","shell.execute_reply":"2023-10-04T08:15:16.709974Z"},"trusted":true},"execution_count":125,"outputs":[]},{"cell_type":"code","source":"# average number of words or tokens in training data\nround(sum([len(i.split()) for i in train_sentences])/len(train_sentences))","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:16.714251Z","iopub.execute_input":"2023-10-04T08:15:16.714983Z","iopub.status.idle":"2023-10-04T08:15:16.735327Z","shell.execute_reply.started":"2023-10-04T08:15:16.714950Z","shell.execute_reply":"2023-10-04T08:15:16.734271Z"},"trusted":true},"execution_count":126,"outputs":[{"execution_count":126,"output_type":"execute_result","data":{"text/plain":"15"},"metadata":{}}]},{"cell_type":"code","source":"text_vectorizer = TextVectorization(max_tokens = 10000,\n                                    output_mode = \"int\",\n                                    output_sequence_length = 15)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:16.736648Z","iopub.execute_input":"2023-10-04T08:15:16.737051Z","iopub.status.idle":"2023-10-04T08:15:16.760346Z","shell.execute_reply.started":"2023-10-04T08:15:16.737021Z","shell.execute_reply":"2023-10-04T08:15:16.758827Z"},"trusted":true},"execution_count":127,"outputs":[]},{"cell_type":"code","source":"# fit the text vectorizer to data\ntext_vectorizer.adapt(train_sentences)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:16.762013Z","iopub.execute_input":"2023-10-04T08:15:16.763194Z","iopub.status.idle":"2023-10-04T08:15:17.213876Z","shell.execute_reply.started":"2023-10-04T08:15:16.763160Z","shell.execute_reply":"2023-10-04T08:15:17.212655Z"},"trusted":true},"execution_count":128,"outputs":[]},{"cell_type":"code","source":"# checking for the text vectorizer \nrandom_sentence = random.choice(train_sentences)\nprint(f\"original {random_sentence}\")\nprint(f\"Vectorized: {text_vectorizer([random_sentence])}\")","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:17.215916Z","iopub.execute_input":"2023-10-04T08:15:17.216271Z","iopub.status.idle":"2023-10-04T08:15:17.253754Z","shell.execute_reply.started":"2023-10-04T08:15:17.216239Z","shell.execute_reply":"2023-10-04T08:15:17.252587Z"},"trusted":true},"execution_count":129,"outputs":[{"name":"stdout","text":"original @eggalie haha I love hurricane because of you\nVectorized: [[  1 847   8 110 346 152   6  12   0   0   0   0   0   0   0]]\n","output_type":"stream"}]},{"cell_type":"code","source":"# unique words in our dictionary\nwords_in_vocab = text_vectorizer.get_vocabulary()\ntop_5 = words_in_vocab[:5]\nbottom_5 = words_in_vocab[-5:]\n\nprint(\"Vocab\", len(words_in_vocab))\nprint(\"Top 5 words:\", top_5)\nprint(\"Bottom 5 words:\", bottom_5)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:17.255370Z","iopub.execute_input":"2023-10-04T08:15:17.255680Z","iopub.status.idle":"2023-10-04T08:15:17.282741Z","shell.execute_reply.started":"2023-10-04T08:15:17.255651Z","shell.execute_reply":"2023-10-04T08:15:17.281684Z"},"trusted":true},"execution_count":130,"outputs":[{"name":"stdout","text":"Vocab 10000\nTop 5 words: ['', '[UNK]', 'the', 'a', 'in']\nBottom 5 words: ['pages', 'paeds', 'pads', 'padres', 'paddytomlinson1']\n","output_type":"stream"}]},{"cell_type":"markdown","source":"## Creating Embeddings","metadata":{}},{"cell_type":"code","source":"tf.random.set_seed(42)\nfrom tensorflow.keras import layers\n\nembedding = layers.Embedding(input_dim = 10000,\n                            output_dim = 128,\n                            embeddings_initializer = \"uniform\",\n                            input_length = 15,\n                            name = \"embedding_1\")\nembedding","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:17.283885Z","iopub.execute_input":"2023-10-04T08:15:17.284157Z","iopub.status.idle":"2023-10-04T08:15:17.471391Z","shell.execute_reply.started":"2023-10-04T08:15:17.284132Z","shell.execute_reply":"2023-10-04T08:15:17.470065Z"},"trusted":true},"execution_count":131,"outputs":[{"execution_count":131,"output_type":"execute_result","data":{"text/plain":"<keras.layers.core.embedding.Embedding at 0x7a1a572e5960>"},"metadata":{}}]},{"cell_type":"code","source":"# example from sample data\n\nrandom_sentence = random.choice(train_sentences)\nprint(f\"original {random_sentence}\")\nprint(f\"Vectorized: {text_vectorizer([random_sentence])}\")\nprint(f\"Vectorized: {embedding(text_vectorizer([random_sentence]))}\")","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:17.472931Z","iopub.execute_input":"2023-10-04T08:15:17.473229Z","iopub.status.idle":"2023-10-04T08:15:17.546286Z","shell.execute_reply.started":"2023-10-04T08:15:17.473204Z","shell.execute_reply":"2023-10-04T08:15:17.545172Z"},"trusted":true},"execution_count":132,"outputs":[{"name":"stdout","text":"original I need an arcade shooter fix but CTE is empty &amp; only running obliteration. I'd even buy a CoD title if they weren't all overpriced on steam\nVectorized: [[   8  162   39    1 4635 2113   30    1    9 5731   35  126  876  536\n   508]]\nVectorized: [[[ 0.01519376 -0.0233181   0.04915943 ... -0.04084576 -0.04359012\n    0.01152353]\n  [ 0.04639918  0.00330198 -0.04080838 ...  0.00779966 -0.01707214\n   -0.02896821]\n  [ 0.01898116 -0.00657962  0.00129762 ... -0.01735419  0.04049825\n    0.03328992]\n  ...\n  [-0.03819742  0.03403808  0.04688605 ...  0.04019666 -0.01951964\n    0.04713443]\n  [ 0.02788652 -0.02451069  0.03905913 ... -0.02860194 -0.02970216\n    0.04306314]\n  [ 0.02649759 -0.00278244 -0.04084604 ... -0.03511374  0.00136449\n    0.03049216]]]\n","output_type":"stream"}]},{"cell_type":"markdown","source":"## Creating Our Models","metadata":{}},{"cell_type":"markdown","source":"### 0. Naive Bayes - Baseline","metadata":{}},{"cell_type":"code","source":"from sklearn.feature_extraction.text import TfidfVectorizer\nfrom sklearn.naive_bayes import MultinomialNB\nfrom sklearn.pipeline import Pipeline\n\n# create tokenizartiona and modelling the pipeline\nmodel_0 = Pipeline([\n    \n    (\"tfdf\", TfidfVectorizer()),\n    (\"clf\", MultinomialNB())\n])\n\nmodel_0.fit(train_sentences, train_labels)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:17.547914Z","iopub.execute_input":"2023-10-04T08:15:17.548193Z","iopub.status.idle":"2023-10-04T08:15:17.734422Z","shell.execute_reply.started":"2023-10-04T08:15:17.548168Z","shell.execute_reply":"2023-10-04T08:15:17.733257Z"},"trusted":true},"execution_count":133,"outputs":[{"execution_count":133,"output_type":"execute_result","data":{"text/plain":"Pipeline(steps=[('tfdf', TfidfVectorizer()), ('clf', MultinomialNB())])","text/html":"<style>#sk-container-id-2 {color: black;background-color: white;}#sk-container-id-2 pre{padding: 0;}#sk-container-id-2 div.sk-toggleable {background-color: white;}#sk-container-id-2 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-2 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-2 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-2 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-2 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-2 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-2 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-2 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-2 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-2 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-2 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-2 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-2 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-2 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-2 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-2 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-2 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-2 div.sk-item {position: relative;z-index: 1;}#sk-container-id-2 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-2 div.sk-item::before, #sk-container-id-2 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-2 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-2 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-2 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-2 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-2 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-2 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-2 div.sk-label-container {text-align: center;}#sk-container-id-2 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-2 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-2\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>Pipeline(steps=[(&#x27;tfdf&#x27;, TfidfVectorizer()), (&#x27;clf&#x27;, MultinomialNB())])</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-4\" type=\"checkbox\" ><label for=\"sk-estimator-id-4\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[(&#x27;tfdf&#x27;, TfidfVectorizer()), (&#x27;clf&#x27;, MultinomialNB())])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-5\" type=\"checkbox\" ><label for=\"sk-estimator-id-5\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">TfidfVectorizer</label><div class=\"sk-toggleable__content\"><pre>TfidfVectorizer()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-6\" type=\"checkbox\" ><label for=\"sk-estimator-id-6\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">MultinomialNB</label><div class=\"sk-toggleable__content\"><pre>MultinomialNB()</pre></div></div></div></div></div></div></div>"},"metadata":{}}]},{"cell_type":"code","source":"baseline_score = model_0.score(val_sentences, val_labels)\nbaseline_score","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:17.735517Z","iopub.execute_input":"2023-10-04T08:15:17.735786Z","iopub.status.idle":"2023-10-04T08:15:17.759542Z","shell.execute_reply.started":"2023-10-04T08:15:17.735762Z","shell.execute_reply":"2023-10-04T08:15:17.758358Z"},"trusted":true},"execution_count":134,"outputs":[{"execution_count":134,"output_type":"execute_result","data":{"text/plain":"0.7926509186351706"},"metadata":{}}]},{"cell_type":"code","source":"baseline_preds = model_0.predict(val_sentences)\nbaseline_preds[:20]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:17.761022Z","iopub.execute_input":"2023-10-04T08:15:17.761471Z","iopub.status.idle":"2023-10-04T08:15:17.786901Z","shell.execute_reply.started":"2023-10-04T08:15:17.761430Z","shell.execute_reply":"2023-10-04T08:15:17.785855Z"},"trusted":true},"execution_count":135,"outputs":[{"execution_count":135,"output_type":"execute_result","data":{"text/plain":"array([1, 1, 1, 0, 0, 1, 1, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1])"},"metadata":{}}]},{"cell_type":"markdown","source":"#### Creating Evaluation functions for our future use","metadata":{}},{"cell_type":"code","source":"from sklearn.metrics import accuracy_score, precision_recall_fscore_support\n\ndef calculate_results(y_true, y_pred):\n    \n    # accuracy\n    model_accuracy = accuracy_score(y_true, y_pred) * 100\n    \n    # model_precision,f1_score, recall using weighted average\n    model_precision, model_recall, model_f1, _ = precision_recall_fscore_support(y_true, y_pred, average=\"weighted\")\n    \n    model_results = {\n        \n        \"accurracy\": model_accuracy,\n        \"precision\": model_precision,\n        \"recall\": model_recall,\n        \"f1\": model_f1\n    }\n    \n    return model_results","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:17.788086Z","iopub.execute_input":"2023-10-04T08:15:17.788382Z","iopub.status.idle":"2023-10-04T08:15:17.794319Z","shell.execute_reply.started":"2023-10-04T08:15:17.788357Z","shell.execute_reply":"2023-10-04T08:15:17.793227Z"},"trusted":true},"execution_count":136,"outputs":[]},{"cell_type":"code","source":"baseline_results = calculate_results(y_true = val_labels,\n                                     y_pred = baseline_preds)\nbaseline_results","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:17.795637Z","iopub.execute_input":"2023-10-04T08:15:17.796029Z","iopub.status.idle":"2023-10-04T08:15:17.815752Z","shell.execute_reply.started":"2023-10-04T08:15:17.795971Z","shell.execute_reply":"2023-10-04T08:15:17.814660Z"},"trusted":true},"execution_count":137,"outputs":[{"execution_count":137,"output_type":"execute_result","data":{"text/plain":"{'accurracy': 79.26509186351706,\n 'precision': 0.8111390004213173,\n 'recall': 0.7926509186351706,\n 'f1': 0.7862189758049549}"},"metadata":{}}]},{"cell_type":"markdown","source":"### Model 1 - Simple Dense","metadata":{}},{"cell_type":"code","source":"# importing helper functions\n!wget https://raw.githubusercontent.com/mrdbourke/tensorflow-deep-learning/main/extras/helper_functions.py","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:17.817272Z","iopub.execute_input":"2023-10-04T08:15:17.817564Z","iopub.status.idle":"2023-10-04T08:15:19.278050Z","shell.execute_reply.started":"2023-10-04T08:15:17.817536Z","shell.execute_reply":"2023-10-04T08:15:19.276652Z"},"trusted":true},"execution_count":138,"outputs":[{"name":"stdout","text":"--2023-10-04 08:15:19--  https://raw.githubusercontent.com/mrdbourke/tensorflow-deep-learning/main/extras/helper_functions.py\nResolving raw.githubusercontent.com (raw.githubusercontent.com)... 185.199.108.133, 185.199.109.133, 185.199.110.133, ...\nConnecting to raw.githubusercontent.com (raw.githubusercontent.com)|185.199.108.133|:443... connected.\nHTTP request sent, awaiting response... 200 OK\nLength: 10246 (10K) [text/plain]\nSaving to: ‘helper_functions.py.1’\n\nhelper_functions.py 100%[===================>]  10.01K  --.-KB/s    in 0s      \n\n2023-10-04 08:15:19 (55.8 MB/s) - ‘helper_functions.py.1’ saved [10246/10246]\n\n","output_type":"stream"}]},{"cell_type":"code","source":"from helper_functions import unzip_data, create_tensorboard_callback, plot_loss_curves, compare_historys","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:19.279862Z","iopub.execute_input":"2023-10-04T08:15:19.280225Z","iopub.status.idle":"2023-10-04T08:15:19.286125Z","shell.execute_reply.started":"2023-10-04T08:15:19.280191Z","shell.execute_reply":"2023-10-04T08:15:19.284860Z"},"trusted":true},"execution_count":139,"outputs":[]},{"cell_type":"code","source":"# create directory to save log \nSAVE_DIR = \"model_logs\"","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:19.287717Z","iopub.execute_input":"2023-10-04T08:15:19.288619Z","iopub.status.idle":"2023-10-04T08:15:19.301124Z","shell.execute_reply.started":"2023-10-04T08:15:19.288580Z","shell.execute_reply":"2023-10-04T08:15:19.299890Z"},"trusted":true},"execution_count":140,"outputs":[]},{"cell_type":"code","source":"# building the model \ninputs = layers.Input(shape=(1,), dtype = \"string\")\nx = text_vectorizer(inputs)\nx = embedding(x)\nx = layers.GlobalAveragePooling1D()(x)\noutputs = layers.Dense(1, activation = \"sigmoid\")(x)\nmodel_1 = tf.keras.Model(inputs, outputs, name = \"model_1_choice\")","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:19.302932Z","iopub.execute_input":"2023-10-04T08:15:19.303627Z","iopub.status.idle":"2023-10-04T08:15:19.399636Z","shell.execute_reply.started":"2023-10-04T08:15:19.303593Z","shell.execute_reply":"2023-10-04T08:15:19.398232Z"},"trusted":true},"execution_count":141,"outputs":[]},{"cell_type":"code","source":"# compile the model\nmodel_1.compile(loss = \"binary_crossentropy\",\n                optimizer = tf.keras.optimizers.Adam(),\n                metrics = [\"accuracy\"])","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:19.401637Z","iopub.execute_input":"2023-10-04T08:15:19.402093Z","iopub.status.idle":"2023-10-04T08:15:19.420831Z","shell.execute_reply.started":"2023-10-04T08:15:19.402046Z","shell.execute_reply":"2023-10-04T08:15:19.419333Z"},"trusted":true},"execution_count":142,"outputs":[]},{"cell_type":"code","source":"model_1.summary()","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:19.421922Z","iopub.execute_input":"2023-10-04T08:15:19.422663Z","iopub.status.idle":"2023-10-04T08:15:19.450357Z","shell.execute_reply.started":"2023-10-04T08:15:19.422631Z","shell.execute_reply":"2023-10-04T08:15:19.449156Z"},"trusted":true},"execution_count":143,"outputs":[{"name":"stdout","text":"Model: \"model_1_choice\"\n_________________________________________________________________\n Layer (type)                Output Shape              Param #   \n=================================================================\n input_6 (InputLayer)        [(None, 1)]               0         \n                                                                 \n text_vectorization_1 (TextV  (None, 15)               0         \n ectorization)                                                   \n                                                                 \n embedding_1 (Embedding)     (None, 15, 128)           1280000   \n                                                                 \n global_average_pooling1d_2   (None, 128)              0         \n (GlobalAveragePooling1D)                                        \n                                                                 \n dense_7 (Dense)             (None, 1)                 129       \n                                                                 \n=================================================================\nTotal params: 1,280,129\nTrainable params: 1,280,129\nNon-trainable params: 0\n_________________________________________________________________\n","output_type":"stream"}]},{"cell_type":"code","source":"# fit the model \nmodel_1_history = model_1.fit(train_sentences,\n                              train_labels,\n                              epochs = 5,\n                              validation_data = (val_sentences, val_labels),\n                              callbacks = [create_tensorboard_callback(dir_name=SAVE_DIR, \n                                                                     experiment_name=\"simple_dense_model\")])","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:19.457764Z","iopub.execute_input":"2023-10-04T08:15:19.458133Z","iopub.status.idle":"2023-10-04T08:15:34.969399Z","shell.execute_reply.started":"2023-10-04T08:15:19.458103Z","shell.execute_reply":"2023-10-04T08:15:34.968420Z"},"trusted":true},"execution_count":144,"outputs":[{"name":"stdout","text":"Saving TensorBoard log files to: model_logs/simple_dense_model/20231004-081519\nEpoch 1/5\n215/215 [==============================] - 4s 15ms/step - loss: 0.6093 - accuracy: 0.6958 - val_loss: 0.5361 - val_accuracy: 0.7559\nEpoch 2/5\n215/215 [==============================] - 3s 13ms/step - loss: 0.4418 - accuracy: 0.8206 - val_loss: 0.4692 - val_accuracy: 0.7861\nEpoch 3/5\n215/215 [==============================] - 3s 13ms/step - loss: 0.3470 - accuracy: 0.8609 - val_loss: 0.4588 - val_accuracy: 0.7927\nEpoch 4/5\n215/215 [==============================] - 3s 13ms/step - loss: 0.2854 - accuracy: 0.8920 - val_loss: 0.4637 - val_accuracy: 0.7927\nEpoch 5/5\n215/215 [==============================] - 3s 13ms/step - loss: 0.2384 - accuracy: 0.9126 - val_loss: 0.4760 - val_accuracy: 0.7887\n","output_type":"stream"}]},{"cell_type":"code","source":"model_1.evaluate(val_sentences, val_labels)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:34.970601Z","iopub.execute_input":"2023-10-04T08:15:34.971196Z","iopub.status.idle":"2023-10-04T08:15:35.101861Z","shell.execute_reply.started":"2023-10-04T08:15:34.971156Z","shell.execute_reply":"2023-10-04T08:15:35.100879Z"},"trusted":true},"execution_count":145,"outputs":[{"name":"stdout","text":"24/24 [==============================] - 0s 2ms/step - loss: 0.4760 - accuracy: 0.7887\n","output_type":"stream"},{"execution_count":145,"output_type":"execute_result","data":{"text/plain":"[0.47604769468307495, 0.7887139320373535]"},"metadata":{}}]},{"cell_type":"code","source":"embedding.weights","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:35.103009Z","iopub.execute_input":"2023-10-04T08:15:35.103282Z","iopub.status.idle":"2023-10-04T08:15:35.113765Z","shell.execute_reply.started":"2023-10-04T08:15:35.103257Z","shell.execute_reply":"2023-10-04T08:15:35.112885Z"},"trusted":true},"execution_count":146,"outputs":[{"execution_count":146,"output_type":"execute_result","data":{"text/plain":"[<tf.Variable 'embedding_1/embeddings:0' shape=(10000, 128) dtype=float32, numpy=\n array([[-0.01596312, -0.02734495,  0.01050767, ..., -0.05528403,\n          0.00534431, -0.00810083],\n        [ 0.00367453,  0.01745513, -0.03607399, ..., -0.04037447,\n          0.03302921, -0.00250512],\n        [-0.01737417,  0.00057014,  0.02603259, ...,  0.02105457,\n         -0.05616429,  0.0582607 ],\n        ...,\n        [-0.01442311, -0.0326    , -0.04227994, ...,  0.0491709 ,\n         -0.03354257,  0.03110572],\n        [-0.05857125, -0.02137076, -0.00058275, ..., -0.07276617,\n         -0.00477383,  0.01959108],\n        [-0.1157446 , -0.09909692,  0.10486263, ..., -0.05489131,\n         -0.05916625,  0.05572179]], dtype=float32)>]"},"metadata":{}}]},{"cell_type":"code","source":"embed_weights = model_1.get_layer(\"embedding_1\").get_weights()[0]\nprint(embed_weights.shape)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:35.115466Z","iopub.execute_input":"2023-10-04T08:15:35.116174Z","iopub.status.idle":"2023-10-04T08:15:35.127165Z","shell.execute_reply.started":"2023-10-04T08:15:35.116135Z","shell.execute_reply":"2023-10-04T08:15:35.126323Z"},"trusted":true},"execution_count":147,"outputs":[{"name":"stdout","text":"(10000, 128)\n","output_type":"stream"}]},{"cell_type":"code","source":"# make predictions\nmodel_1_preds = tf.squeeze(tf.round(model_1.predict(val_sentences)))\nmodel_1_preds[:10]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:35.128405Z","iopub.execute_input":"2023-10-04T08:15:35.129146Z","iopub.status.idle":"2023-10-04T08:15:35.348172Z","shell.execute_reply.started":"2023-10-04T08:15:35.129117Z","shell.execute_reply":"2023-10-04T08:15:35.346856Z"},"trusted":true},"execution_count":148,"outputs":[{"name":"stdout","text":"24/24 [==============================] - 0s 2ms/step\n","output_type":"stream"},{"execution_count":148,"output_type":"execute_result","data":{"text/plain":"<tf.Tensor: shape=(10,), dtype=float32, numpy=array([0., 1., 1., 0., 0., 1., 1., 1., 1., 0.], dtype=float32)>"},"metadata":{}}]},{"cell_type":"code","source":"# model_1 metrics \nmodel_1_results = calculate_results(y_true = val_labels,\n                                    y_pred = model_1_preds)\n\nmodel_1_results","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:35.349966Z","iopub.execute_input":"2023-10-04T08:15:35.350298Z","iopub.status.idle":"2023-10-04T08:15:35.362867Z","shell.execute_reply.started":"2023-10-04T08:15:35.350263Z","shell.execute_reply":"2023-10-04T08:15:35.361628Z"},"trusted":true},"execution_count":149,"outputs":[{"execution_count":149,"output_type":"execute_result","data":{"text/plain":"{'accurracy': 78.87139107611549,\n 'precision': 0.7930566476728028,\n 'recall': 0.7887139107611548,\n 'f1': 0.7859560656123432}"},"metadata":{}}]},{"cell_type":"code","source":"# comparing results of baseline with new models \n\ndef compare_with_baseline(baseline_results, new_model_results):\n    for key, value in baseline_results.items():\n        print(f\"Baseline {key}: {value:.2f}, New {key}: {new_model_results[key]:.2f}, Difference: {new_model_results[key] - value:.2f}\")\n","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:35.364488Z","iopub.execute_input":"2023-10-04T08:15:35.364959Z","iopub.status.idle":"2023-10-04T08:15:35.371042Z","shell.execute_reply.started":"2023-10-04T08:15:35.364891Z","shell.execute_reply":"2023-10-04T08:15:35.369902Z"},"trusted":true},"execution_count":150,"outputs":[]},{"cell_type":"code","source":"compare_with_baseline(baseline_results=baseline_results, \n                                new_model_results=model_1_results)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:35.372453Z","iopub.execute_input":"2023-10-04T08:15:35.372767Z","iopub.status.idle":"2023-10-04T08:15:35.386002Z","shell.execute_reply.started":"2023-10-04T08:15:35.372738Z","shell.execute_reply":"2023-10-04T08:15:35.385009Z"},"trusted":true},"execution_count":151,"outputs":[{"name":"stdout","text":"Baseline accurracy: 79.27, New accurracy: 78.87, Difference: -0.39\nBaseline precision: 0.81, New precision: 0.79, Difference: -0.02\nBaseline recall: 0.79, New recall: 0.79, Difference: -0.00\nBaseline f1: 0.79, New f1: 0.79, Difference: -0.00\n","output_type":"stream"}]},{"cell_type":"markdown","source":"## Model 2 - LSTM ","metadata":{}},{"cell_type":"code","source":"tf.random.set_seed(42)\n\nmodel_2_embedding = layers.Embedding(input_dim = 10000,\n                                     output_dim = 128,\n                                     embeddings_initializer = \"uniform\",\n                                     input_length = 15,\n                                     name = \"embedding_2\")\n\n# LSTM Model\ninputs = layers.Input(shape = (1,), dtype = \"string\")\nx = text_vectorizer(inputs)\nx = model_2_embedding(x)\nprint(x.shape)\nx = layers.LSTM(64)(x)\nprint(x.shape)\noutputs = layers.Dense(1, activation = \"sigmoid\")(x)\nmodel_2 = tf.keras.Model(inputs, outputs , name = \"model_2_LSTM\")","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:35.387476Z","iopub.execute_input":"2023-10-04T08:15:35.387896Z","iopub.status.idle":"2023-10-04T08:15:35.797974Z","shell.execute_reply.started":"2023-10-04T08:15:35.387856Z","shell.execute_reply":"2023-10-04T08:15:35.796731Z"},"trusted":true},"execution_count":152,"outputs":[{"name":"stdout","text":"(None, 15, 128)\n(None, 64)\n","output_type":"stream"}]},{"cell_type":"code","source":"# compile the model \n\nmodel_2.compile(loss = \"binary_crossentropy\",\n                optimizer = tf.keras.optimizers.Adam(),\n                metrics = [\"accuracy\"])\n\nmodel_2.summary()","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:35.799207Z","iopub.execute_input":"2023-10-04T08:15:35.800101Z","iopub.status.idle":"2023-10-04T08:15:35.832509Z","shell.execute_reply.started":"2023-10-04T08:15:35.800070Z","shell.execute_reply":"2023-10-04T08:15:35.831486Z"},"trusted":true},"execution_count":153,"outputs":[{"name":"stdout","text":"Model: \"model_2_LSTM\"\n_________________________________________________________________\n Layer (type)                Output Shape              Param #   \n=================================================================\n input_7 (InputLayer)        [(None, 1)]               0         \n                                                                 \n text_vectorization_1 (TextV  (None, 15)               0         \n ectorization)                                                   \n                                                                 \n embedding_2 (Embedding)     (None, 15, 128)           1280000   \n                                                                 \n lstm_2 (LSTM)               (None, 64)                49408     \n                                                                 \n dense_8 (Dense)             (None, 1)                 65        \n                                                                 \n=================================================================\nTotal params: 1,329,473\nTrainable params: 1,329,473\nNon-trainable params: 0\n_________________________________________________________________\n","output_type":"stream"}]},{"cell_type":"code","source":"# fit the model \nmodel_2_history = model_2.fit(train_sentences,\n                              train_labels,\n                              epochs = 5,\n                              validation_data = (val_sentences, val_labels),\n                              callbacks = [create_tensorboard_callback(SAVE_DIR, \n                                                                     \"LSTM\")])","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:15:35.834015Z","iopub.execute_input":"2023-10-04T08:15:35.834288Z","iopub.status.idle":"2023-10-04T08:16:03.465599Z","shell.execute_reply.started":"2023-10-04T08:15:35.834263Z","shell.execute_reply":"2023-10-04T08:16:03.464353Z"},"trusted":true},"execution_count":154,"outputs":[{"name":"stdout","text":"Saving TensorBoard log files to: model_logs/LSTM/20231004-081535\nEpoch 1/5\n215/215 [==============================] - 8s 26ms/step - loss: 0.5088 - accuracy: 0.7462 - val_loss: 0.4558 - val_accuracy: 0.7743\nEpoch 2/5\n215/215 [==============================] - 5s 23ms/step - loss: 0.3169 - accuracy: 0.8705 - val_loss: 0.5113 - val_accuracy: 0.7808\nEpoch 3/5\n215/215 [==============================] - 5s 23ms/step - loss: 0.2196 - accuracy: 0.9169 - val_loss: 0.5870 - val_accuracy: 0.7598\nEpoch 4/5\n215/215 [==============================] - 5s 22ms/step - loss: 0.1552 - accuracy: 0.9448 - val_loss: 0.6298 - val_accuracy: 0.7703\nEpoch 5/5\n215/215 [==============================] - 5s 23ms/step - loss: 0.1079 - accuracy: 0.9596 - val_loss: 0.8189 - val_accuracy: 0.7730\n","output_type":"stream"}]},{"cell_type":"code","source":"model_2_pred_probs = model_2.predict(val_sentences)\nmodel_2_pred_probs.shape, model_2_pred_probs[:10]\n","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:16:03.467228Z","iopub.execute_input":"2023-10-04T08:16:03.467523Z","iopub.status.idle":"2023-10-04T08:16:04.109080Z","shell.execute_reply.started":"2023-10-04T08:16:03.467496Z","shell.execute_reply":"2023-10-04T08:16:04.108043Z"},"trusted":true},"execution_count":155,"outputs":[{"name":"stdout","text":"24/24 [==============================] - 1s 5ms/step\n","output_type":"stream"},{"execution_count":155,"output_type":"execute_result","data":{"text/plain":"((762, 1),\n array([[0.01673127],\n        [0.75856405],\n        [0.9994714 ],\n        [0.08984704],\n        [0.00456853],\n        [0.99941266],\n        [0.91473925],\n        [0.99971545],\n        [0.9994867 ],\n        [0.26380113]], dtype=float32))"},"metadata":{}}]},{"cell_type":"code","source":"model_2_preds = tf.squeeze(tf.round(model_2_pred_probs))\nmodel_2_preds[:10]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:16:04.110322Z","iopub.execute_input":"2023-10-04T08:16:04.110614Z","iopub.status.idle":"2023-10-04T08:16:04.121688Z","shell.execute_reply.started":"2023-10-04T08:16:04.110586Z","shell.execute_reply":"2023-10-04T08:16:04.120727Z"},"trusted":true},"execution_count":156,"outputs":[{"execution_count":156,"output_type":"execute_result","data":{"text/plain":"<tf.Tensor: shape=(10,), dtype=float32, numpy=array([0., 1., 1., 0., 0., 1., 1., 1., 1., 0.], dtype=float32)>"},"metadata":{}}]},{"cell_type":"code","source":"model_2_results = calculate_results(y_true=val_labels,\n                                    y_pred=model_2_preds)\nmodel_2_results","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:16:04.123394Z","iopub.execute_input":"2023-10-04T08:16:04.123859Z","iopub.status.idle":"2023-10-04T08:16:04.138760Z","shell.execute_reply.started":"2023-10-04T08:16:04.123815Z","shell.execute_reply":"2023-10-04T08:16:04.137384Z"},"trusted":true},"execution_count":157,"outputs":[{"execution_count":157,"output_type":"execute_result","data":{"text/plain":"{'accurracy': 77.29658792650919,\n 'precision': 0.775649278370019,\n 'recall': 0.7729658792650919,\n 'f1': 0.7704381702262061}"},"metadata":{}}]},{"cell_type":"code","source":"compare_with_baseline(baseline_results, model_2_results)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:16:04.140699Z","iopub.execute_input":"2023-10-04T08:16:04.141219Z","iopub.status.idle":"2023-10-04T08:16:04.148365Z","shell.execute_reply.started":"2023-10-04T08:16:04.141174Z","shell.execute_reply":"2023-10-04T08:16:04.146909Z"},"trusted":true},"execution_count":158,"outputs":[{"name":"stdout","text":"Baseline accurracy: 79.27, New accurracy: 77.30, Difference: -1.97\nBaseline precision: 0.81, New precision: 0.78, Difference: -0.04\nBaseline recall: 0.79, New recall: 0.77, Difference: -0.02\nBaseline f1: 0.79, New f1: 0.77, Difference: -0.02\n","output_type":"stream"}]},{"cell_type":"markdown","source":"### Model 3 -- GRU","metadata":{}},{"cell_type":"code","source":"model_3_embedding = layers.Embedding(input_dim = 10000,\n                                     output_dim = 128,\n                                     embeddings_initializer = \"uniform\",\n                                     input_length = 15,\n                                     name = \"embedding_3\")\n\n# model \ninputs = layers.Input(shape = (1,), dtype = 'string')\nx = text_vectorizer(inputs)\nx = model_3_embedding(x)\nx = layers.GRU(64)(x)\noutputs = layers.Dense(1, activation = \"sigmoid\")(x)\n\nmodel_3 = tf.keras.Model(inputs, outputs, name = \"model_3_GRU\")","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:16:04.150253Z","iopub.execute_input":"2023-10-04T08:16:04.150590Z","iopub.status.idle":"2023-10-04T08:16:04.469841Z","shell.execute_reply.started":"2023-10-04T08:16:04.150558Z","shell.execute_reply":"2023-10-04T08:16:04.468629Z"},"trusted":true},"execution_count":159,"outputs":[]},{"cell_type":"code","source":"model_3.compile( loss = \"binary_crossentropy\",\n                 optimizer = tf.keras.optimizers.Adam(),\n                 metrics = [\"accuracy\"])","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:16:04.471002Z","iopub.execute_input":"2023-10-04T08:16:04.471319Z","iopub.status.idle":"2023-10-04T08:16:04.485429Z","shell.execute_reply.started":"2023-10-04T08:16:04.471292Z","shell.execute_reply":"2023-10-04T08:16:04.484195Z"},"trusted":true},"execution_count":160,"outputs":[]},{"cell_type":"code","source":"model_3.summary()","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:16:04.486439Z","iopub.execute_input":"2023-10-04T08:16:04.486747Z","iopub.status.idle":"2023-10-04T08:16:04.513598Z","shell.execute_reply.started":"2023-10-04T08:16:04.486720Z","shell.execute_reply":"2023-10-04T08:16:04.512545Z"},"trusted":true},"execution_count":161,"outputs":[{"name":"stdout","text":"Model: \"model_3_GRU\"\n_________________________________________________________________\n Layer (type)                Output Shape              Param #   \n=================================================================\n input_8 (InputLayer)        [(None, 1)]               0         \n                                                                 \n text_vectorization_1 (TextV  (None, 15)               0         \n ectorization)                                                   \n                                                                 \n embedding_3 (Embedding)     (None, 15, 128)           1280000   \n                                                                 \n gru_1 (GRU)                 (None, 64)                37248     \n                                                                 \n dense_9 (Dense)             (None, 1)                 65        \n                                                                 \n=================================================================\nTotal params: 1,317,313\nTrainable params: 1,317,313\nNon-trainable params: 0\n_________________________________________________________________\n","output_type":"stream"}]},{"cell_type":"code","source":"model_3_history = model_3.fit(train_sentences,\n                              train_labels, \n                              epochs = 5,\n                              validation_data = (val_sentences, val_labels),\n                              callbacks=[create_tensorboard_callback(SAVE_DIR, \"GRU\")])","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:16:04.515476Z","iopub.execute_input":"2023-10-04T08:16:04.515860Z","iopub.status.idle":"2023-10-04T08:16:47.741600Z","shell.execute_reply.started":"2023-10-04T08:16:04.515823Z","shell.execute_reply":"2023-10-04T08:16:47.740428Z"},"trusted":true},"execution_count":162,"outputs":[{"name":"stdout","text":"Saving TensorBoard log files to: model_logs/GRU/20231004-081604\nEpoch 1/5\n215/215 [==============================] - 8s 26ms/step - loss: 0.5294 - accuracy: 0.7240 - val_loss: 0.4554 - val_accuracy: 0.7743\nEpoch 2/5\n215/215 [==============================] - 5s 24ms/step - loss: 0.3197 - accuracy: 0.8705 - val_loss: 0.4876 - val_accuracy: 0.7782\nEpoch 3/5\n215/215 [==============================] - 5s 22ms/step - loss: 0.2158 - accuracy: 0.9190 - val_loss: 0.5572 - val_accuracy: 0.7677\nEpoch 4/5\n215/215 [==============================] - 5s 22ms/step - loss: 0.1516 - accuracy: 0.9469 - val_loss: 0.6275 - val_accuracy: 0.7769\nEpoch 5/5\n215/215 [==============================] - 5s 22ms/step - loss: 0.1146 - accuracy: 0.9607 - val_loss: 0.5995 - val_accuracy: 0.7703\n","output_type":"stream"}]},{"cell_type":"code","source":"model_3_pred_probs = model_3.predict(val_sentences)\nmodel_3_pred_probs.shape, model_3_pred_probs[:10]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:16:47.743196Z","iopub.execute_input":"2023-10-04T08:16:47.743541Z","iopub.status.idle":"2023-10-04T08:16:48.468712Z","shell.execute_reply.started":"2023-10-04T08:16:47.743509Z","shell.execute_reply":"2023-10-04T08:16:48.467434Z"},"trusted":true},"execution_count":163,"outputs":[{"name":"stdout","text":"24/24 [==============================] - 1s 4ms/step\n","output_type":"stream"},{"execution_count":163,"output_type":"execute_result","data":{"text/plain":"((762, 1),\n array([[0.2949066 ],\n        [0.93698865],\n        [0.9975806 ],\n        [0.18327007],\n        [0.01145282],\n        [0.99118376],\n        [0.7835394 ],\n        [0.99683845],\n        [0.99524844],\n        [0.19951855]], dtype=float32))"},"metadata":{}}]},{"cell_type":"code","source":"model_3_preds = tf.squeeze(tf.round(model_3_pred_probs))\nmodel_3_preds[:10]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:16:48.470335Z","iopub.execute_input":"2023-10-04T08:16:48.471233Z","iopub.status.idle":"2023-10-04T08:16:48.481416Z","shell.execute_reply.started":"2023-10-04T08:16:48.471187Z","shell.execute_reply":"2023-10-04T08:16:48.480051Z"},"trusted":true},"execution_count":164,"outputs":[{"execution_count":164,"output_type":"execute_result","data":{"text/plain":"<tf.Tensor: shape=(10,), dtype=float32, numpy=array([0., 1., 1., 0., 0., 1., 1., 1., 1., 0.], dtype=float32)>"},"metadata":{}}]},{"cell_type":"code","source":"model_3_results = calculate_results(y_true=val_labels, \n                                    y_pred=model_3_preds)\nmodel_3_results","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:16:48.483040Z","iopub.execute_input":"2023-10-04T08:16:48.484160Z","iopub.status.idle":"2023-10-04T08:16:48.501265Z","shell.execute_reply.started":"2023-10-04T08:16:48.484122Z","shell.execute_reply":"2023-10-04T08:16:48.500079Z"},"trusted":true},"execution_count":165,"outputs":[{"execution_count":165,"output_type":"execute_result","data":{"text/plain":"{'accurracy': 77.03412073490814,\n 'precision': 0.7702878929185921,\n 'recall': 0.7703412073490814,\n 'f1': 0.769333028556703}"},"metadata":{}}]},{"cell_type":"code","source":"compare_with_baseline(baseline_results, model_3_results)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:16:48.502771Z","iopub.execute_input":"2023-10-04T08:16:48.503296Z","iopub.status.idle":"2023-10-04T08:16:48.511005Z","shell.execute_reply.started":"2023-10-04T08:16:48.503264Z","shell.execute_reply":"2023-10-04T08:16:48.509977Z"},"trusted":true},"execution_count":166,"outputs":[{"name":"stdout","text":"Baseline accurracy: 79.27, New accurracy: 77.03, Difference: -2.23\nBaseline precision: 0.81, New precision: 0.77, Difference: -0.04\nBaseline recall: 0.79, New recall: 0.77, Difference: -0.02\nBaseline f1: 0.79, New f1: 0.77, Difference: -0.02\n","output_type":"stream"}]},{"cell_type":"markdown","source":"### Bidirection RNN ","metadata":{}},{"cell_type":"code","source":"model_4_embedding = layers.Embedding(input_dim = 10000,\n                                     output_dim = 128,\n                                     embeddings_initializer=\"uniform\",\n                                     input_length = 15,\n                                     name = \"embedding_4\")\n\n# model\n\ninputs = layers.Input(shape=(1,), dtype = \"string\")\nx = text_vectorizer(inputs)\nx = model_4_embedding(x)\nx = layers.Bidirectional(layers.LSTM(64))(x)\noutputs = layers.Dense(1, activation = \"sigmoid\")(x)\n\nmodel_4 = tf.keras.Model(inputs, outputs, name = \"model_4_Bidrectional\")\nmodel_4.summary()","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:16:48.512321Z","iopub.execute_input":"2023-10-04T08:16:48.512976Z","iopub.status.idle":"2023-10-04T08:16:49.128604Z","shell.execute_reply.started":"2023-10-04T08:16:48.512944Z","shell.execute_reply":"2023-10-04T08:16:49.127465Z"},"trusted":true},"execution_count":167,"outputs":[{"name":"stdout","text":"Model: \"model_4_Bidrectional\"\n_________________________________________________________________\n Layer (type)                Output Shape              Param #   \n=================================================================\n input_9 (InputLayer)        [(None, 1)]               0         \n                                                                 \n text_vectorization_1 (TextV  (None, 15)               0         \n ectorization)                                                   \n                                                                 \n embedding_4 (Embedding)     (None, 15, 128)           1280000   \n                                                                 \n bidirectional_1 (Bidirectio  (None, 128)              98816     \n nal)                                                            \n                                                                 \n dense_10 (Dense)            (None, 1)                 129       \n                                                                 \n=================================================================\nTotal params: 1,378,945\nTrainable params: 1,378,945\nNon-trainable params: 0\n_________________________________________________________________\n","output_type":"stream"}]},{"cell_type":"code","source":"model_4.compile( loss = \"binary_crossentropy\",\n                 optimizer = tf.keras.optimizers.Adam(),\n                 metrics = [\"accuracy\"])","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:16:49.130265Z","iopub.execute_input":"2023-10-04T08:16:49.130583Z","iopub.status.idle":"2023-10-04T08:16:49.144136Z","shell.execute_reply.started":"2023-10-04T08:16:49.130553Z","shell.execute_reply":"2023-10-04T08:16:49.142993Z"},"trusted":true},"execution_count":168,"outputs":[]},{"cell_type":"code","source":"model_4_history = model_4.fit(train_sentences,\n                              train_labels, \n                              epochs = 5,\n                              validation_data = (val_sentences, val_labels),\n                              callbacks=[create_tensorboard_callback(SAVE_DIR, \"bidirectional_RNN\")])","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:16:49.146065Z","iopub.execute_input":"2023-10-04T08:16:49.146491Z","iopub.status.idle":"2023-10-04T08:17:22.048759Z","shell.execute_reply.started":"2023-10-04T08:16:49.146446Z","shell.execute_reply":"2023-10-04T08:17:22.047925Z"},"trusted":true},"execution_count":169,"outputs":[{"name":"stdout","text":"Saving TensorBoard log files to: model_logs/bidirectional_RNN/20231004-081649\nEpoch 1/5\n215/215 [==============================] - 11s 29ms/step - loss: 0.5093 - accuracy: 0.7473 - val_loss: 0.4592 - val_accuracy: 0.7822\nEpoch 2/5\n215/215 [==============================] - 5s 24ms/step - loss: 0.3140 - accuracy: 0.8718 - val_loss: 0.5128 - val_accuracy: 0.7756\nEpoch 3/5\n215/215 [==============================] - 6s 26ms/step - loss: 0.2145 - accuracy: 0.9169 - val_loss: 0.5689 - val_accuracy: 0.7703\nEpoch 4/5\n215/215 [==============================] - 6s 27ms/step - loss: 0.1502 - accuracy: 0.9495 - val_loss: 0.6367 - val_accuracy: 0.7743\nEpoch 5/5\n215/215 [==============================] - 5s 24ms/step - loss: 0.1054 - accuracy: 0.9634 - val_loss: 0.6968 - val_accuracy: 0.7638\n","output_type":"stream"}]},{"cell_type":"code","source":"model_4_pred_probs = model_4.predict(val_sentences)\nmodel_4_pred_probs[:10]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:17:22.050545Z","iopub.execute_input":"2023-10-04T08:17:22.051074Z","iopub.status.idle":"2023-10-04T08:17:23.021285Z","shell.execute_reply.started":"2023-10-04T08:17:22.051039Z","shell.execute_reply":"2023-10-04T08:17:23.019885Z"},"trusted":true},"execution_count":170,"outputs":[{"name":"stdout","text":"24/24 [==============================] - 1s 6ms/step\n","output_type":"stream"},{"execution_count":170,"output_type":"execute_result","data":{"text/plain":"array([[0.11629684],\n       [0.921876  ],\n       [0.9996061 ],\n       [0.24089041],\n       [0.00810372],\n       [0.99763894],\n       [0.98037297],\n       [0.9995756 ],\n       [0.9997297 ],\n       [0.2959498 ]], dtype=float32)"},"metadata":{}}]},{"cell_type":"code","source":"model_4_preds = tf.squeeze(tf.round(model_4_pred_probs))\nmodel_4_preds[:10]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:17:23.022855Z","iopub.execute_input":"2023-10-04T08:17:23.023204Z","iopub.status.idle":"2023-10-04T08:17:23.035957Z","shell.execute_reply.started":"2023-10-04T08:17:23.023173Z","shell.execute_reply":"2023-10-04T08:17:23.034635Z"},"trusted":true},"execution_count":171,"outputs":[{"execution_count":171,"output_type":"execute_result","data":{"text/plain":"<tf.Tensor: shape=(10,), dtype=float32, numpy=array([0., 1., 1., 0., 0., 1., 1., 1., 1., 0.], dtype=float32)>"},"metadata":{}}]},{"cell_type":"code","source":"model_4_results = calculate_results(val_labels, model_4_preds)\nmodel_4_results","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:17:23.037308Z","iopub.execute_input":"2023-10-04T08:17:23.037649Z","iopub.status.idle":"2023-10-04T08:17:23.057210Z","shell.execute_reply.started":"2023-10-04T08:17:23.037620Z","shell.execute_reply":"2023-10-04T08:17:23.056012Z"},"trusted":true},"execution_count":172,"outputs":[{"execution_count":172,"output_type":"execute_result","data":{"text/plain":"{'accurracy': 76.37795275590551,\n 'precision': 0.7635200205237502,\n 'recall': 0.7637795275590551,\n 'f1': 0.7636021959563469}"},"metadata":{}}]},{"cell_type":"code","source":"compare_with_baseline(baseline_results, model_4_results)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:17:23.058385Z","iopub.execute_input":"2023-10-04T08:17:23.058826Z","iopub.status.idle":"2023-10-04T08:17:23.068927Z","shell.execute_reply.started":"2023-10-04T08:17:23.058766Z","shell.execute_reply":"2023-10-04T08:17:23.067553Z"},"trusted":true},"execution_count":173,"outputs":[{"name":"stdout","text":"Baseline accurracy: 79.27, New accurracy: 76.38, Difference: -2.89\nBaseline precision: 0.81, New precision: 0.76, Difference: -0.05\nBaseline recall: 0.79, New recall: 0.76, Difference: -0.03\nBaseline f1: 0.79, New f1: 0.76, Difference: -0.02\n","output_type":"stream"}]},{"cell_type":"markdown","source":"### CNN1D","metadata":{}},{"cell_type":"code","source":"model_5_embedding = layers.Embedding(input_dim = 10000,\n                                     output_dim = 128,\n                                     embeddings_initializer = \"uniform\",\n                                     input_length = 15,\n                                     name = \"embedding_5\")\n\n\n# model \ninputs = layers.Input(shape=(1,), dtype = \"string\")\nx = text_vectorizer(inputs)\nx = model_5_embedding(x)\nx = layers.Conv1D(filters = 32, kernel_size = 5, activation = \"relu\")(x)\nx = layers.GlobalAveragePooling1D()(x)\noutputs = layers.Dense(1, activation = \"sigmoid\")(x)\n\nmodel_5 = tf.keras.Model(inputs, outputs, name=\"model_5_Conv1D\")\nmodel_5.summary()\n","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:17:23.070093Z","iopub.execute_input":"2023-10-04T08:17:23.070413Z","iopub.status.idle":"2023-10-04T08:17:23.201967Z","shell.execute_reply.started":"2023-10-04T08:17:23.070384Z","shell.execute_reply":"2023-10-04T08:17:23.200523Z"},"trusted":true},"execution_count":174,"outputs":[{"name":"stdout","text":"Model: \"model_5_Conv1D\"\n_________________________________________________________________\n Layer (type)                Output Shape              Param #   \n=================================================================\n input_10 (InputLayer)       [(None, 1)]               0         \n                                                                 \n text_vectorization_1 (TextV  (None, 15)               0         \n ectorization)                                                   \n                                                                 \n embedding_5 (Embedding)     (None, 15, 128)           1280000   \n                                                                 \n conv1d_1 (Conv1D)           (None, 11, 32)            20512     \n                                                                 \n global_average_pooling1d_3   (None, 32)               0         \n (GlobalAveragePooling1D)                                        \n                                                                 \n dense_11 (Dense)            (None, 1)                 33        \n                                                                 \n=================================================================\nTotal params: 1,300,545\nTrainable params: 1,300,545\nNon-trainable params: 0\n_________________________________________________________________\n","output_type":"stream"}]},{"cell_type":"code","source":"model_5.compile(loss = \"binary_crossentropy\",\n                optimizer = tf.keras.optimizers.Adam(),\n                metrics = [\"accuracy\"])","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:17:23.203622Z","iopub.execute_input":"2023-10-04T08:17:23.203978Z","iopub.status.idle":"2023-10-04T08:17:23.216505Z","shell.execute_reply.started":"2023-10-04T08:17:23.203949Z","shell.execute_reply":"2023-10-04T08:17:23.215295Z"},"trusted":true},"execution_count":175,"outputs":[]},{"cell_type":"code","source":"model_5_history = model_5.fit(train_sentences,\n                              train_labels,\n                              epochs = 5,\n                              validation_data = (val_sentences, val_labels),\n                              callbacks = [create_tensorboard_callback(SAVE_DIR, \n                                                                     \"Conv1D\")])","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:17:23.218858Z","iopub.execute_input":"2023-10-04T08:17:23.219665Z","iopub.status.idle":"2023-10-04T08:17:39.551097Z","shell.execute_reply.started":"2023-10-04T08:17:23.219612Z","shell.execute_reply":"2023-10-04T08:17:39.549898Z"},"trusted":true},"execution_count":176,"outputs":[{"name":"stdout","text":"Saving TensorBoard log files to: model_logs/Conv1D/20231004-081723\nEpoch 1/5\n215/215 [==============================] - 4s 15ms/step - loss: 0.5361 - accuracy: 0.7342 - val_loss: 0.4667 - val_accuracy: 0.7822\nEpoch 2/5\n215/215 [==============================] - 3s 15ms/step - loss: 0.3374 - accuracy: 0.8577 - val_loss: 0.5084 - val_accuracy: 0.7835\nEpoch 3/5\n215/215 [==============================] - 3s 14ms/step - loss: 0.2329 - accuracy: 0.9105 - val_loss: 0.5804 - val_accuracy: 0.7664\nEpoch 4/5\n215/215 [==============================] - 3s 14ms/step - loss: 0.1684 - accuracy: 0.9348 - val_loss: 0.6753 - val_accuracy: 0.7730\nEpoch 5/5\n215/215 [==============================] - 3s 14ms/step - loss: 0.1259 - accuracy: 0.9511 - val_loss: 0.7693 - val_accuracy: 0.7559\n","output_type":"stream"}]},{"cell_type":"code","source":"model_5_pred_probs = model_5.predict(val_sentences)\nmodel_5_pred_probs[:10]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:17:39.552842Z","iopub.execute_input":"2023-10-04T08:17:39.553199Z","iopub.status.idle":"2023-10-04T08:17:39.776682Z","shell.execute_reply.started":"2023-10-04T08:17:39.553170Z","shell.execute_reply":"2023-10-04T08:17:39.775500Z"},"trusted":true},"execution_count":177,"outputs":[{"name":"stdout","text":"24/24 [==============================] - 0s 2ms/step\n","output_type":"stream"},{"execution_count":177,"output_type":"execute_result","data":{"text/plain":"array([[0.53268176],\n       [0.8252157 ],\n       [0.9999978 ],\n       [0.019157  ],\n       [0.01490563],\n       [0.9856388 ],\n       [0.8462774 ],\n       [0.9996877 ],\n       [0.9999983 ],\n       [0.15805149]], dtype=float32)"},"metadata":{}}]},{"cell_type":"code","source":"model_5_preds = tf.squeeze(tf.round(model_5_pred_probs))\nmodel_5_preds[:10]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:17:39.778291Z","iopub.execute_input":"2023-10-04T08:17:39.779064Z","iopub.status.idle":"2023-10-04T08:17:39.789171Z","shell.execute_reply.started":"2023-10-04T08:17:39.779032Z","shell.execute_reply":"2023-10-04T08:17:39.788097Z"},"trusted":true},"execution_count":178,"outputs":[{"execution_count":178,"output_type":"execute_result","data":{"text/plain":"<tf.Tensor: shape=(10,), dtype=float32, numpy=array([1., 1., 1., 0., 0., 1., 1., 1., 1., 0.], dtype=float32)>"},"metadata":{}}]},{"cell_type":"code","source":"model_5_results = calculate_results(y_true=val_labels, \n                                    y_pred=model_5_preds)\nmodel_5_results","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:17:39.790648Z","iopub.execute_input":"2023-10-04T08:17:39.791045Z","iopub.status.idle":"2023-10-04T08:17:39.803753Z","shell.execute_reply.started":"2023-10-04T08:17:39.790998Z","shell.execute_reply":"2023-10-04T08:17:39.802416Z"},"trusted":true},"execution_count":179,"outputs":[{"execution_count":179,"output_type":"execute_result","data":{"text/plain":"{'accurracy': 75.59055118110236,\n 'precision': 0.7555819221227483,\n 'recall': 0.7559055118110236,\n 'f1': 0.7549846322026649}"},"metadata":{}}]},{"cell_type":"code","source":"compare_with_baseline(baseline_results, model_5_results)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:17:39.805591Z","iopub.execute_input":"2023-10-04T08:17:39.806296Z","iopub.status.idle":"2023-10-04T08:17:39.815432Z","shell.execute_reply.started":"2023-10-04T08:17:39.806258Z","shell.execute_reply":"2023-10-04T08:17:39.814656Z"},"trusted":true},"execution_count":180,"outputs":[{"name":"stdout","text":"Baseline accurracy: 79.27, New accurracy: 75.59, Difference: -3.67\nBaseline precision: 0.81, New precision: 0.76, Difference: -0.06\nBaseline recall: 0.79, New recall: 0.76, Difference: -0.04\nBaseline f1: 0.79, New f1: 0.75, Difference: -0.03\n","output_type":"stream"}]},{"cell_type":"markdown","source":"### Model 6: TensorFlow Hub Pretrained Sentence Encoder","metadata":{}},{"cell_type":"code","source":"import tensorflow_hub as hub","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:17:39.816783Z","iopub.execute_input":"2023-10-04T08:17:39.817418Z","iopub.status.idle":"2023-10-04T08:17:39.828721Z","shell.execute_reply.started":"2023-10-04T08:17:39.817389Z","shell.execute_reply":"2023-10-04T08:17:39.827626Z"},"trusted":true},"execution_count":181,"outputs":[]},{"cell_type":"code","source":"sentence_encoder_layer = hub.KerasLayer(\"https://tfhub.dev/google/universal-sentence-encoder/4\",\n                                        input_shape=[],\n                                        dtype=tf.string,\n                                        trainable=False,\n                                        name=\"USE\")","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:17:39.830166Z","iopub.execute_input":"2023-10-04T08:17:39.830979Z","iopub.status.idle":"2023-10-04T08:17:45.500474Z","shell.execute_reply.started":"2023-10-04T08:17:39.830928Z","shell.execute_reply":"2023-10-04T08:17:45.499272Z"},"trusted":true},"execution_count":182,"outputs":[]},{"cell_type":"code","source":"model_6 = tf.keras.Sequential([\n    \n    sentence_encoder_layer,\n    layers.Dense(64, activation = \"relu\"),\n    layers.Dense(1, activation = \"relu\"),\n], name = 'model_6')\n\nmodel_6.compile(loss = \"binary_crossentropy\",\n                optimizer = tf.keras.optimizers.Adam(),\n                metrics = [\"accuracy\"])\n\nmodel_6.summary()","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:17:45.501770Z","iopub.execute_input":"2023-10-04T08:17:45.502093Z","iopub.status.idle":"2023-10-04T08:17:45.738701Z","shell.execute_reply.started":"2023-10-04T08:17:45.502056Z","shell.execute_reply":"2023-10-04T08:17:45.737645Z"},"trusted":true},"execution_count":183,"outputs":[{"name":"stdout","text":"Model: \"model_6\"\n_________________________________________________________________\n Layer (type)                Output Shape              Param #   \n=================================================================\n USE (KerasLayer)            (None, 512)               256797824 \n                                                                 \n dense_12 (Dense)            (None, 64)                32832     \n                                                                 \n dense_13 (Dense)            (None, 1)                 65        \n                                                                 \n=================================================================\nTotal params: 256,830,721\nTrainable params: 32,897\nNon-trainable params: 256,797,824\n_________________________________________________________________\n","output_type":"stream"}]},{"cell_type":"code","source":"model_6_history = model_6.fit(train_sentences,\n                              train_labels,\n                              epochs = 5,\n                              validation_data = (val_sentences, val_labels),\n                              callbacks = [create_tensorboard_callback(SAVE_DIR, \n                                                                     \"tf_hub_sentence_encoder\")])","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:17:45.740423Z","iopub.execute_input":"2023-10-04T08:17:45.741137Z","iopub.status.idle":"2023-10-04T08:18:08.988581Z","shell.execute_reply.started":"2023-10-04T08:17:45.741096Z","shell.execute_reply":"2023-10-04T08:18:08.987550Z"},"trusted":true},"execution_count":184,"outputs":[{"name":"stdout","text":"Saving TensorBoard log files to: model_logs/tf_hub_sentence_encoder/20231004-081745\nEpoch 1/5\n215/215 [==============================] - 6s 12ms/step - loss: 0.5422 - accuracy: 0.7688 - val_loss: 0.4545 - val_accuracy: 0.7900\nEpoch 2/5\n215/215 [==============================] - 2s 10ms/step - loss: 0.4288 - accuracy: 0.8149 - val_loss: 0.4667 - val_accuracy: 0.8136\nEpoch 3/5\n215/215 [==============================] - 2s 10ms/step - loss: 0.4117 - accuracy: 0.8222 - val_loss: 0.5043 - val_accuracy: 0.8097\nEpoch 4/5\n215/215 [==============================] - 2s 11ms/step - loss: 0.3996 - accuracy: 0.8289 - val_loss: 0.5254 - val_accuracy: 0.8084\nEpoch 5/5\n215/215 [==============================] - 2s 11ms/step - loss: 0.3851 - accuracy: 0.8278 - val_loss: 0.6111 - val_accuracy: 0.8163\n","output_type":"stream"}]},{"cell_type":"code","source":"model_6_pred_probs = model_6.predict(val_sentences)\nmodel_6_pred_probs[:10]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:08.990360Z","iopub.execute_input":"2023-10-04T08:18:08.990781Z","iopub.status.idle":"2023-10-04T08:18:09.649133Z","shell.execute_reply.started":"2023-10-04T08:18:08.990739Z","shell.execute_reply":"2023-10-04T08:18:09.647992Z"},"trusted":true},"execution_count":185,"outputs":[{"name":"stdout","text":"24/24 [==============================] - 1s 8ms/step\n","output_type":"stream"},{"execution_count":185,"output_type":"execute_result","data":{"text/plain":"array([[0.21527252],\n       [0.71246314],\n       [1.2832165 ],\n       [0.08408416],\n       [0.69894516],\n       [0.7037981 ],\n       [1.3153126 ],\n       [1.2945757 ],\n       [0.9245631 ],\n       [0.20188162]], dtype=float32)"},"metadata":{}}]},{"cell_type":"code","source":"model_6_preds = tf.squeeze(tf.round(model_6_pred_probs))\nmodel_6_preds[:10]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:09.650686Z","iopub.execute_input":"2023-10-04T08:18:09.651158Z","iopub.status.idle":"2023-10-04T08:18:09.662066Z","shell.execute_reply.started":"2023-10-04T08:18:09.651117Z","shell.execute_reply":"2023-10-04T08:18:09.660721Z"},"trusted":true},"execution_count":186,"outputs":[{"execution_count":186,"output_type":"execute_result","data":{"text/plain":"<tf.Tensor: shape=(10,), dtype=float32, numpy=array([0., 1., 1., 0., 1., 1., 1., 1., 1., 0.], dtype=float32)>"},"metadata":{}}]},{"cell_type":"code","source":"model_6_results = calculate_results(val_labels, model_6_preds)\nmodel_6_results","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:09.663567Z","iopub.execute_input":"2023-10-04T08:18:09.664126Z","iopub.status.idle":"2023-10-04T08:18:09.678635Z","shell.execute_reply.started":"2023-10-04T08:18:09.664092Z","shell.execute_reply":"2023-10-04T08:18:09.677700Z"},"trusted":true},"execution_count":187,"outputs":[{"name":"stderr","text":"/opt/conda/lib/python3.10/site-packages/sklearn/metrics/_classification.py:1344: UndefinedMetricWarning: Recall and F-score are ill-defined and being set to 0.0 in labels with no true samples. Use `zero_division` parameter to control this behavior.\n  _warn_prf(average, modifier, msg_start, len(result))\n","output_type":"stream"},{"execution_count":187,"output_type":"execute_result","data":{"text/plain":"{'accurracy': 81.10236220472441,\n 'precision': 0.8201136427095549,\n 'recall': 0.8110236220472441,\n 'f1': 0.8105537221917096}"},"metadata":{}}]},{"cell_type":"code","source":"compare_with_baseline(baseline_results, model_6_results)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:09.680189Z","iopub.execute_input":"2023-10-04T08:18:09.680844Z","iopub.status.idle":"2023-10-04T08:18:09.688012Z","shell.execute_reply.started":"2023-10-04T08:18:09.680779Z","shell.execute_reply":"2023-10-04T08:18:09.687079Z"},"trusted":true},"execution_count":188,"outputs":[{"name":"stdout","text":"Baseline accurracy: 79.27, New accurracy: 81.10, Difference: 1.84\nBaseline precision: 0.81, New precision: 0.82, Difference: 0.01\nBaseline recall: 0.79, New recall: 0.81, Difference: 0.02\nBaseline f1: 0.79, New f1: 0.81, Difference: 0.02\n","output_type":"stream"}]},{"cell_type":"markdown","source":"### Model 7: TensorFlow Hub Pretrained Sentence Encoder 10% of the training data","metadata":{}},{"cell_type":"code","source":"# split our data into 10% training\nimport numpy as np\ntrain_90, train_10, train_labels_90, train_labels_10 = train_test_split(np.array(train_sentences),\n                                                                        train_labels,\n                                                                        test_size=0.1,\n                                                                        random_state=42)\n   \n                                           ","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:09.689321Z","iopub.execute_input":"2023-10-04T08:18:09.690312Z","iopub.status.idle":"2023-10-04T08:18:09.704293Z","shell.execute_reply.started":"2023-10-04T08:18:09.690260Z","shell.execute_reply":"2023-10-04T08:18:09.703127Z"},"trusted":true},"execution_count":189,"outputs":[]},{"cell_type":"code","source":"len(train_sentences), len(train_10)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:09.705781Z","iopub.execute_input":"2023-10-04T08:18:09.706136Z","iopub.status.idle":"2023-10-04T08:18:09.726622Z","shell.execute_reply.started":"2023-10-04T08:18:09.706104Z","shell.execute_reply":"2023-10-04T08:18:09.725703Z"},"trusted":true},"execution_count":190,"outputs":[{"execution_count":190,"output_type":"execute_result","data":{"text/plain":"(6851, 686)"},"metadata":{}}]},{"cell_type":"code","source":"pd.Series(train_labels_10).value_counts()","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:09.728385Z","iopub.execute_input":"2023-10-04T08:18:09.729139Z","iopub.status.idle":"2023-10-04T08:18:09.742581Z","shell.execute_reply.started":"2023-10-04T08:18:09.729095Z","shell.execute_reply":"2023-10-04T08:18:09.741175Z"},"trusted":true},"execution_count":191,"outputs":[{"execution_count":191,"output_type":"execute_result","data":{"text/plain":"0    415\n1    271\nName: count, dtype: int64"},"metadata":{}}]},{"cell_type":"code","source":"# model 7 \nmodel_7 = tf.keras.models.clone_model(model_6)\n\nmodel_7.compile(loss = \"binary_crossentropy\",\n                optimizer = \"Adam\",\n                metrics = [\"accuracy\"])\n\nmodel_7.summary()","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:09.744063Z","iopub.execute_input":"2023-10-04T08:18:09.744361Z","iopub.status.idle":"2023-10-04T08:18:15.780821Z","shell.execute_reply.started":"2023-10-04T08:18:09.744335Z","shell.execute_reply":"2023-10-04T08:18:15.779553Z"},"trusted":true},"execution_count":192,"outputs":[{"name":"stdout","text":"Model: \"model_6\"\n_________________________________________________________________\n Layer (type)                Output Shape              Param #   \n=================================================================\n USE (KerasLayer)            (None, 512)               256797824 \n                                                                 \n dense_12 (Dense)            (None, 64)                32832     \n                                                                 \n dense_13 (Dense)            (None, 1)                 65        \n                                                                 \n=================================================================\nTotal params: 256,830,721\nTrainable params: 32,897\nNon-trainable params: 256,797,824\n_________________________________________________________________\n","output_type":"stream"}]},{"cell_type":"code","source":"model_7_history = model_7.fit(x=train_10,\n                              y=train_labels_10,\n                              epochs=5,\n                              validation_data=(val_sentences, val_labels),\n                              callbacks=[create_tensorboard_callback(SAVE_DIR, \"10_percent_tf_hub_sentence_encoder\")])","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:15.782187Z","iopub.execute_input":"2023-10-04T08:18:15.782492Z","iopub.status.idle":"2023-10-04T08:18:21.628061Z","shell.execute_reply.started":"2023-10-04T08:18:15.782466Z","shell.execute_reply":"2023-10-04T08:18:21.626873Z"},"trusted":true},"execution_count":193,"outputs":[{"name":"stdout","text":"Saving TensorBoard log files to: model_logs/10_percent_tf_hub_sentence_encoder/20231004-081815\nEpoch 1/5\n22/22 [==============================] - 4s 47ms/step - loss: 1.2548 - accuracy: 0.6152 - val_loss: 0.6126 - val_accuracy: 0.6378\nEpoch 2/5\n22/22 [==============================] - 1s 25ms/step - loss: 0.5134 - accuracy: 0.7784 - val_loss: 0.5215 - val_accuracy: 0.7861\nEpoch 3/5\n22/22 [==============================] - 0s 19ms/step - loss: 0.4530 - accuracy: 0.8222 - val_loss: 0.5058 - val_accuracy: 0.7979\nEpoch 4/5\n22/22 [==============================] - 1s 24ms/step - loss: 0.4020 - accuracy: 0.8382 - val_loss: 0.5101 - val_accuracy: 0.7940\nEpoch 5/5\n22/22 [==============================] - 0s 19ms/step - loss: 0.3646 - accuracy: 0.8367 - val_loss: 0.5956 - val_accuracy: 0.7966\n","output_type":"stream"}]},{"cell_type":"code","source":"model_7_pred_probs = model_7.predict(val_sentences)\nmodel_7_pred_probs[:10]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:21.629454Z","iopub.execute_input":"2023-10-04T08:18:21.630325Z","iopub.status.idle":"2023-10-04T08:18:22.250534Z","shell.execute_reply.started":"2023-10-04T08:18:21.630293Z","shell.execute_reply":"2023-10-04T08:18:22.249514Z"},"trusted":true},"execution_count":194,"outputs":[{"name":"stdout","text":"24/24 [==============================] - 1s 8ms/step\n","output_type":"stream"},{"execution_count":194,"output_type":"execute_result","data":{"text/plain":"array([[0.26882705],\n       [0.7708482 ],\n       [1.0354844 ],\n       [0.11638144],\n       [0.62312514],\n       [0.86240566],\n       [0.7969029 ],\n       [0.9477657 ],\n       [0.90370923],\n       [0.03552004]], dtype=float32)"},"metadata":{}}]},{"cell_type":"code","source":"model_7_preds = tf.squeeze(tf.round(model_7_pred_probs))\nmodel_7_preds[:10]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:22.251983Z","iopub.execute_input":"2023-10-04T08:18:22.253142Z","iopub.status.idle":"2023-10-04T08:18:22.262900Z","shell.execute_reply.started":"2023-10-04T08:18:22.253096Z","shell.execute_reply":"2023-10-04T08:18:22.261653Z"},"trusted":true},"execution_count":195,"outputs":[{"execution_count":195,"output_type":"execute_result","data":{"text/plain":"<tf.Tensor: shape=(10,), dtype=float32, numpy=array([0., 1., 1., 0., 1., 1., 1., 1., 1., 0.], dtype=float32)>"},"metadata":{}}]},{"cell_type":"code","source":"model_7_results = calculate_results(val_labels, model_7_preds)\nmodel_7_results","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:22.264531Z","iopub.execute_input":"2023-10-04T08:18:22.264963Z","iopub.status.idle":"2023-10-04T08:18:22.281163Z","shell.execute_reply.started":"2023-10-04T08:18:22.264922Z","shell.execute_reply":"2023-10-04T08:18:22.279648Z"},"trusted":true},"execution_count":196,"outputs":[{"execution_count":196,"output_type":"execute_result","data":{"text/plain":"{'accurracy': 79.65879265091863,\n 'precision': 0.8041691930196172,\n 'recall': 0.7965879265091863,\n 'f1': 0.7930665218571817}"},"metadata":{}}]},{"cell_type":"code","source":"compare_with_baseline(baseline_results, model_7_results)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:22.282691Z","iopub.execute_input":"2023-10-04T08:18:22.283283Z","iopub.status.idle":"2023-10-04T08:18:22.291681Z","shell.execute_reply.started":"2023-10-04T08:18:22.283249Z","shell.execute_reply":"2023-10-04T08:18:22.290177Z"},"trusted":true},"execution_count":197,"outputs":[{"name":"stdout","text":"Baseline accurracy: 79.27, New accurracy: 79.66, Difference: 0.39\nBaseline precision: 0.81, New precision: 0.80, Difference: -0.01\nBaseline recall: 0.79, New recall: 0.80, Difference: 0.00\nBaseline f1: 0.79, New f1: 0.79, Difference: 0.01\n","output_type":"stream"}]},{"cell_type":"markdown","source":"## Comparing Results Of All Models","metadata":{}},{"cell_type":"code","source":"all_model_results = pd.DataFrame({\n    \"Baseline\": baseline_results,\n    'Simple Dense': model_1_results,\n    \"LSTM\": model_2_results,\n    \"GRU\": model_3_results,\n    \"Bidirectional\": model_4_results,\n    \"Conv1D\": model_5_results,\n    \"TF Sentence Encoder\": model_6_results,\n    \"TF 10 Percent Data\": model_7_results,\n})\n\nall_model_results = all_model_results.transpose()\nall_model_results","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:22.302857Z","iopub.execute_input":"2023-10-04T08:18:22.303902Z","iopub.status.idle":"2023-10-04T08:18:22.320425Z","shell.execute_reply.started":"2023-10-04T08:18:22.303849Z","shell.execute_reply":"2023-10-04T08:18:22.319389Z"},"trusted":true},"execution_count":198,"outputs":[{"execution_count":198,"output_type":"execute_result","data":{"text/plain":"                     accurracy  precision    recall        f1\nBaseline             79.265092   0.811139  0.792651  0.786219\nSimple Dense         78.871391   0.793057  0.788714  0.785956\nLSTM                 77.296588   0.775649  0.772966  0.770438\nGRU                  77.034121   0.770288  0.770341  0.769333\nBidirectional        76.377953   0.763520  0.763780  0.763602\nConv1D               75.590551   0.755582  0.755906  0.754985\nTF Sentence Encoder  81.102362   0.820114  0.811024  0.810554\nTF 10 Percent Data   79.658793   0.804169  0.796588  0.793067","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>accurracy</th>\n      <th>precision</th>\n      <th>recall</th>\n      <th>f1</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>Baseline</th>\n      <td>79.265092</td>\n      <td>0.811139</td>\n      <td>0.792651</td>\n      <td>0.786219</td>\n    </tr>\n    <tr>\n      <th>Simple Dense</th>\n      <td>78.871391</td>\n      <td>0.793057</td>\n      <td>0.788714</td>\n      <td>0.785956</td>\n    </tr>\n    <tr>\n      <th>LSTM</th>\n      <td>77.296588</td>\n      <td>0.775649</td>\n      <td>0.772966</td>\n      <td>0.770438</td>\n    </tr>\n    <tr>\n      <th>GRU</th>\n      <td>77.034121</td>\n      <td>0.770288</td>\n      <td>0.770341</td>\n      <td>0.769333</td>\n    </tr>\n    <tr>\n      <th>Bidirectional</th>\n      <td>76.377953</td>\n      <td>0.763520</td>\n      <td>0.763780</td>\n      <td>0.763602</td>\n    </tr>\n    <tr>\n      <th>Conv1D</th>\n      <td>75.590551</td>\n      <td>0.755582</td>\n      <td>0.755906</td>\n      <td>0.754985</td>\n    </tr>\n    <tr>\n      <th>TF Sentence Encoder</th>\n      <td>81.102362</td>\n      <td>0.820114</td>\n      <td>0.811024</td>\n      <td>0.810554</td>\n    </tr>\n    <tr>\n      <th>TF 10 Percent Data</th>\n      <td>79.658793</td>\n      <td>0.804169</td>\n      <td>0.796588</td>\n      <td>0.793067</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"all_model_results[\"accurracy\"] = all_model_results[\"accurracy\"]/100","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:22.321620Z","iopub.execute_input":"2023-10-04T08:18:22.322635Z","iopub.status.idle":"2023-10-04T08:18:22.337006Z","shell.execute_reply.started":"2023-10-04T08:18:22.322602Z","shell.execute_reply":"2023-10-04T08:18:22.335636Z"},"trusted":true},"execution_count":199,"outputs":[]},{"cell_type":"code","source":"all_model_results.plot(kind=\"bar\", figsize=(10, 7)).legend(bbox_to_anchor=(1.0, 1.0));","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:22.338211Z","iopub.execute_input":"2023-10-04T08:18:22.338934Z","iopub.status.idle":"2023-10-04T08:18:22.759373Z","shell.execute_reply.started":"2023-10-04T08:18:22.338902Z","shell.execute_reply":"2023-10-04T08:18:22.758017Z"},"trusted":true},"execution_count":200,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1000x700 with 1 Axes>","image/png":"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"},"metadata":{}}]},{"cell_type":"code","source":"all_model_results.sort_values(\"f1\", ascending=False)[\"f1\"].plot(kind=\"bar\", figsize=(10, 7));","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:22.760724Z","iopub.execute_input":"2023-10-04T08:18:22.761141Z","iopub.status.idle":"2023-10-04T08:18:23.050468Z","shell.execute_reply.started":"2023-10-04T08:18:22.761067Z","shell.execute_reply":"2023-10-04T08:18:23.049386Z"},"trusted":true},"execution_count":201,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1000x700 with 1 Axes>","image/png":"iVBORw0KGgoAAAANSUhEUgAAAzoAAALLCAYAAAAxCPe8AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8pXeV/AAAACXBIWXMAAA9hAAAPYQGoP6dpAABZIklEQVR4nO3deVxU9eL/8TeDgqKIGoEbiaipqLiRSl2XjHLL0pZramlktql5o0wtlbQS65aSN39xc8ls0/TSqtk1XMott8AF9w00wS1BMCEYfn/4bYoAE2PmeD/zej4e83g0n3Nm5g0TMm8+53yOR2FhYaEAAAAAwCA2qwMAAAAAQHmj6AAAAAAwDkUHAAAAgHEoOgAAAACMQ9EBAAAAYByKDgAAAADjUHQAAAAAGKeC1QEuh91u148//ihfX195eHhYHQcAAACARQoLC3Xu3DnVqVNHNlvp8zb/E0Xnxx9/VFBQkNUxAAAAAFwl0tLSVK9evVK3/08UHV9fX0kXv5hq1apZnAYAAACAVbKyshQUFOToCKX5nyg6vx6uVq1aNYoOAAAAgD89pYXFCAAAAAAYh6IDAAAAwDgUHQAAAADGoegAAAAAMA5FBwAAAIBxKDoAAAAAjEPRAQAAAGAcig4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHEoOgAAAACMQ9EBAAAAYByKDgAAAADjUHQAAAAAGIeiAwAAAMA4FB0AAAAAxqHoAAAAADAORQcAAACAcSg6AAAAAIxD0QEAAABgnApWB7haBI9dYnWEv+Tw1N5WRwAAAACuGszoAAAAADAORQcAAACAca6o6MycOVPBwcGqVKmSOnTooI0bN15y/7i4ODVp0kSVK1dWUFCQnnrqKV24cOGKAgMAAADAnylz0Vm4cKGio6MVExOjrVu3qlWrVurevbtOnDhR4v4ffvihxo4dq5iYGO3atUtz5szRwoUL9dxzz/3l8AAAAABQkjIXnWnTpmnYsGGKiopSaGio4uPj5ePjo7lz55a4/7p163TTTTdp4MCBCg4O1m233aYBAwb86SwQAAAAAFypMhWdvLw8bdmyRZGRkb89gc2myMhIrV+/vsTH3HjjjdqyZYuj2Bw8eFBLly5Vr169Sn2d3NxcZWVlFbkBAAAAwOUq0/LSp06dUkFBgQIDA4uMBwYGavfu3SU+ZuDAgTp16pT+9re/qbCwUPn5+XrssccueehabGysJk2aVJZoAAAAAODg9FXXVq1apSlTpuj//b//p61btyohIUFLlizRiy++WOpjxo0bp8zMTMctLS3N2TEBAAAAGKRMMzr+/v7y9PRURkZGkfGMjAzVqlWrxMdMmDBBDzzwgB5++GFJUsuWLZWTk6NHHnlEzz//vGy24l3L29tb3t7eZYkGAAAAAA5lmtHx8vJSu3btlJiY6Biz2+1KTExUREREiY85f/58sTLj6ekpSSosLCxrXgAAAAD4U2Wa0ZGk6OhoDRkyROHh4Wrfvr3i4uKUk5OjqKgoSdLgwYNVt25dxcbGSpL69OmjadOmqU2bNurQoYP279+vCRMmqE+fPo7CA0hS8NglVkf4Sw5P7W11BAAAAPyfMhed/v376+TJk5o4caLS09PVunVrLVu2zLFAQWpqapEZnPHjx8vDw0Pjx4/XsWPHdO2116pPnz56+eWXy++rAAAAAIDf8Sj8Hzh+LCsrS35+fsrMzFS1atWc8hrMJliP9wAAAAB/5nK7gdNXXQMAAAAAVyvzoWsAzMWsGgAAMAVFBwCuIpRNAADKB4euAQAAADAORQcAAACAcSg6AAAAAIxD0QEAAABgHIoOAAAAAONQdAAAAAAYh6IDAAAAwDgUHQAAAADG4YKhAAD8n//1C7ZKXLQVAH5F0QEAAFcNyiaA8sKhawAAAACMQ9EBAAAAYByKDgAAAADjcI4OAAAAHDhPCqZgRgcAAACAcSg6AAAAAIxD0QEAAABgHIoOAAAAAOOwGAEAAABwFWFBiPLBjA4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHEoOgAAAACMQ9EBAAAAYByKDgAAAADjUHQAAAAAGIeiAwAAAMA4FB0AAAAAxqHoAAAAADAORQcAAACAcSg6AAAAAIxD0QEAAABgHIoOAAAAAONQdAAAAAAYh6IDAAAAwDgUHQAAAADGoegAAAAAMA5FBwAAAIBxKDoAAAAAjEPRAQAAAGAcig4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHGuqOjMnDlTwcHBqlSpkjp06KCNGzeWum/Xrl3l4eFR7Na7d+8rDg0AAAAAl1LmorNw4UJFR0crJiZGW7duVatWrdS9e3edOHGixP0TEhJ0/Phxx23Hjh3y9PTUvffe+5fDAwAAAEBJylx0pk2bpmHDhikqKkqhoaGKj4+Xj4+P5s6dW+L+NWvWVK1atRy35cuXy8fHh6IDAAAAwGnKVHTy8vK0ZcsWRUZG/vYENpsiIyO1fv36y3qOOXPm6L777lOVKlVK3Sc3N1dZWVlFbgAAAABwucpUdE6dOqWCggIFBgYWGQ8MDFR6evqfPn7jxo3asWOHHn744UvuFxsbKz8/P8ctKCioLDEBAAAAuDmXrro2Z84ctWzZUu3bt7/kfuPGjVNmZqbjlpaW5qKEAAAAAExQoSw7+/v7y9PTUxkZGUXGMzIyVKtWrUs+NicnRwsWLNDkyZP/9HW8vb3l7e1dlmgAAAAA4FCmGR0vLy+1a9dOiYmJjjG73a7ExERFRERc8rGLFi1Sbm6u7r///itLCgAAAACXqUwzOpIUHR2tIUOGKDw8XO3bt1dcXJxycnIUFRUlSRo8eLDq1q2r2NjYIo+bM2eO+vbtq2uuuaZ8kgMAAABAKcpcdPr376+TJ09q4sSJSk9PV+vWrbVs2TLHAgWpqamy2YpOFO3Zs0dr1qzRf//73/JJDQAAAACXUOaiI0kjRozQiBEjSty2atWqYmNNmjRRYWHhlbwUAAAAAJSZS1ddAwAAAABXoOgAAAAAMA5FBwAAAIBxKDoAAAAAjEPRAQAAAGAcig4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHEoOgAAAACMQ9EBAAAAYByKDgAAAADjUHQAAAAAGIeiAwAAAMA4FB0AAAAAxqHoAAAAADAORQcAAACAcSg6AAAAAIxD0QEAAABgHIoOAAAAAONQdAAAAAAYh6IDAAAAwDgUHQAAAADGoegAAAAAMA5FBwAAAIBxKDoAAAAAjEPRAQAAAGAcig4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHEoOgAAAACMQ9EBAAAAYByKDgAAAADjUHQAAAAAGIeiAwAAAMA4FB0AAAAAxqHoAAAAADAORQcAAACAcSg6AAAAAIxD0QEAAABgHIoOAAAAAONQdAAAAAAYh6IDAAAAwDgUHQAAAADGoegAAAAAMA5FBwAAAIBxKDoAAAAAjEPRAQAAAGCcKyo6M2fOVHBwsCpVqqQOHTpo48aNl9z/7NmzGj58uGrXri1vb29df/31Wrp06RUFBgAAAIA/U6GsD1i4cKGio6MVHx+vDh06KC4uTt27d9eePXsUEBBQbP+8vDzdeuutCggI0OLFi1W3bl0dOXJE1atXL4/8AAAAAFBMmYvOtGnTNGzYMEVFRUmS4uPjtWTJEs2dO1djx44ttv/cuXN15swZrVu3ThUrVpQkBQcHX/I1cnNzlZub67iflZVV1pgAAAAA3FiZDl3Ly8vTli1bFBkZ+dsT2GyKjIzU+vXrS3zM559/roiICA0fPlyBgYFq0aKFpkyZooKCglJfJzY2Vn5+fo5bUFBQWWICAAAAcHNlKjqnTp1SQUGBAgMDi4wHBgYqPT29xMccPHhQixcvVkFBgZYuXaoJEybo9ddf10svvVTq64wbN06ZmZmOW1paWlliAgAAAHBzZT50razsdrsCAgL09ttvy9PTU+3atdOxY8f0z3/+UzExMSU+xtvbW97e3s6OBgAAAMBQZSo6/v7+8vT0VEZGRpHxjIwM1apVq8TH1K5dWxUrVpSnp6djrFmzZkpPT1deXp68vLyuIDYAAAAAlK5Mh655eXmpXbt2SkxMdIzZ7XYlJiYqIiKixMfcdNNN2r9/v+x2u2Ns7969ql27NiUHAAAAgFOU+To60dHRmjVrlt59913t2rVLjz/+uHJychyrsA0ePFjjxo1z7P/444/rzJkzGjVqlPbu3aslS5ZoypQpGj58ePl9FQAAAADwO2U+R6d///46efKkJk6cqPT0dLVu3VrLli1zLFCQmpoqm+23/hQUFKSvv/5aTz31lMLCwlS3bl2NGjVKY8aMKb+vAgAAAAB+54oWIxgxYoRGjBhR4rZVq1YVG4uIiNCGDRuu5KUAAAAAoMzKfOgaAAAAAFztKDoAAAAAjEPRAQAAAGAcig4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHEoOgAAAACMQ9EBAAAAYByKDgAAAADjUHQAAAAAGIeiAwAAAMA4FB0AAAAAxqHoAAAAADAORQcAAACAcSg6AAAAAIxD0QEAAABgHIoOAAAAAONQdAAAAAAYh6IDAAAAwDgUHQAAAADGoegAAAAAMA5FBwAAAIBxKDoAAAAAjEPRAQAAAGAcig4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHEoOgAAAACMQ9EBAAAAYByKDgAAAADjUHQAAAAAGIeiAwAAAMA4FB0AAAAAxqHoAAAAADAORQcAAACAcSg6AAAAAIxD0QEAAABgHIoOAAAAAONQdAAAAAAYh6IDAAAAwDgUHQAAAADGoegAAAAAMA5FBwAAAIBxKDoAAAAAjEPRAQAAAGAcig4AAAAA41xR0Zk5c6aCg4NVqVIldejQQRs3bix133nz5snDw6PIrVKlSlccGAAAAAD+TJmLzsKFCxUdHa2YmBht3bpVrVq1Uvfu3XXixIlSH1OtWjUdP37ccTty5MhfCg0AAAAAl1LmojNt2jQNGzZMUVFRCg0NVXx8vHx8fDR37txSH+Ph4aFatWo5boGBgX8pNAAAAABcSpmKTl5enrZs2aLIyMjfnsBmU2RkpNavX1/q47Kzs1W/fn0FBQXpzjvv1M6dOy/5Orm5ucrKyipyAwAAAIDLVaaic+rUKRUUFBSbkQkMDFR6enqJj2nSpInmzp2rzz77TO+//77sdrtuvPFGHT16tNTXiY2NlZ+fn+MWFBRUlpgAAAAA3JzTV12LiIjQ4MGD1bp1a3Xp0kUJCQm69tpr9e9//7vUx4wbN06ZmZmOW1pamrNjAgAAADBIhbLs7O/vL09PT2VkZBQZz8jIUK1atS7rOSpWrKg2bdpo//79pe7j7e0tb2/vskQDAAAAAIcyzeh4eXmpXbt2SkxMdIzZ7XYlJiYqIiLisp6joKBA27dvV+3atcuWFAAAAAAuU5lmdCQpOjpaQ4YMUXh4uNq3b6+4uDjl5OQoKipKkjR48GDVrVtXsbGxkqTJkyerY8eOatSokc6ePat//vOfOnLkiB5++OHy/UoAAAAA4P+Uuej0799fJ0+e1MSJE5Wenq7WrVtr2bJljgUKUlNTZbP9NlH0008/adiwYUpPT1eNGjXUrl07rVu3TqGhoeX3VQAAAADA75S56EjSiBEjNGLEiBK3rVq1qsj96dOna/r06VfyMgAAAABwRZy+6hoAAAAAuBpFBwAAAIBxKDoAAAAAjEPRAQAAAGAcig4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHEoOgAAAACMQ9EBAAAAYByKDgAAAADjUHQAAAAAGIeiAwAAAMA4FB0AAAAAxqHoAAAAADAORQcAAACAcSg6AAAAAIxD0QEAAABgHIoOAAAAAONQdAAAAAAYh6IDAAAAwDgUHQAAAADGoegAAAAAMA5FBwAAAIBxKDoAAAAAjEPRAQAAAGAcig4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHEoOgAAAACMQ9EBAAAAYByKDgAAAADjUHQAAAAAGIeiAwAAAMA4FB0AAAAAxqHoAAAAADAORQcAAACAcSg6AAAAAIxD0QEAAABgHIoOAAAAAONQdAAAAAAYh6IDAAAAwDgUHQAAAADGoegAAAAAMA5FBwAAAIBxKDoAAAAAjEPRAQAAAGCcKyo6M2fOVHBwsCpVqqQOHTpo48aNl/W4BQsWyMPDQ3379r2SlwUAAACAy1LmorNw4UJFR0crJiZGW7duVatWrdS9e3edOHHiko87fPiwnnnmGXXq1OmKwwIAAADA5Shz0Zk2bZqGDRumqKgohYaGKj4+Xj4+Ppo7d26pjykoKNCgQYM0adIkhYSE/KXAAAAAAPBnylR08vLytGXLFkVGRv72BDabIiMjtX79+lIfN3nyZAUEBGjo0KGX9Tq5ubnKysoqcgMAAACAy1WmonPq1CkVFBQoMDCwyHhgYKDS09NLfMyaNWs0Z84czZo167JfJzY2Vn5+fo5bUFBQWWICAAAAcHNOXXXt3LlzeuCBBzRr1iz5+/tf9uPGjRunzMxMxy0tLc2JKQEAAACYpkJZdvb395enp6cyMjKKjGdkZKhWrVrF9j9w4IAOHz6sPn36OMbsdvvFF65QQXv27FHDhg2LPc7b21ve3t5liQYAAAAADmWa0fHy8lK7du2UmJjoGLPb7UpMTFRERESx/Zs2bart27crKSnJcbvjjjt08803KykpiUPSAAAAADhFmWZ0JCk6OlpDhgxReHi42rdvr7i4OOXk5CgqKkqSNHjwYNWtW1exsbGqVKmSWrRoUeTx1atXl6Ri4wAAAABQXspcdPr376+TJ09q4sSJSk9PV+vWrbVs2TLHAgWpqamy2Zx66g8AAAAAXFKZi44kjRgxQiNGjChx26pVqy752Hnz5l3JSwIAAADAZWPqBQAAAIBxKDoAAAAAjEPRAQAAAGAcig4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHEoOgAAAACMQ9EBAAAAYByKDgAAAADjUHQAAAAAGIeiAwAAAMA4FB0AAAAAxqHoAAAAADAORQcAAACAcSg6AAAAAIxD0QEAAABgHIoOAAAAAONQdAAAAAAYh6IDAAAAwDgUHQAAAADGoegAAAAAMA5FBwAAAIBxKDoAAAAAjEPRAQAAAGAcig4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHEoOgAAAACMQ9EBAAAAYByKDgAAAADjUHQAAAAAGIeiAwAAAMA4FB0AAAAAxqHoAAAAADAORQcAAACAcSg6AAAAAIxD0QEAAABgHIoOAAAAAONQdAAAAAAYh6IDAAAAwDgUHQAAAADGoegAAAAAMA5FBwAAAIBxKDoAAAAAjEPRAQAAAGAcig4AAAAA41B0AAAAABjniorOzJkzFRwcrEqVKqlDhw7auHFjqfsmJCQoPDxc1atXV5UqVdS6dWu99957VxwYAAAAAP5MmYvOwoULFR0drZiYGG3dulWtWrVS9+7ddeLEiRL3r1mzpp5//nmtX79e27ZtU1RUlKKiovT111//5fAAAAAAUJIyF51p06Zp2LBhioqKUmhoqOLj4+Xj46O5c+eWuH/Xrl3Vr18/NWvWTA0bNtSoUaMUFhamNWvWlPoaubm5ysrKKnIDAAAAgMtVpqKTl5enLVu2KDIy8rcnsNkUGRmp9evX/+njCwsLlZiYqD179qhz586l7hcbGys/Pz/HLSgoqCwxAQAAALi5MhWdU6dOqaCgQIGBgUXGAwMDlZ6eXurjMjMzVbVqVXl5eal3797617/+pVtvvbXU/ceNG6fMzEzHLS0trSwxAQAAALi5Cq54EV9fXyUlJSk7O1uJiYmKjo5WSEiIunbtWuL+3t7e8vb2dkU0AAAAAAYqU9Hx9/eXp6enMjIyioxnZGSoVq1apT7OZrOpUaNGkqTWrVtr165dio2NLbXoAAAAAMBfUaZD17y8vNSuXTslJiY6xux2uxITExUREXHZz2O325Wbm1uWlwYAAACAy1bmQ9eio6M1ZMgQhYeHq3379oqLi1NOTo6ioqIkSYMHD1bdunUVGxsr6eLCAuHh4WrYsKFyc3O1dOlSvffee3rrrbfK9ysBAAAAgP9T5qLTv39/nTx5UhMnTlR6erpat26tZcuWORYoSE1Nlc3220RRTk6OnnjiCR09elSVK1dW06ZN9f7776t///7l91UAAAAAwO9c0WIEI0aM0IgRI0rctmrVqiL3X3rpJb300ktX8jIAAAAAcEXKfMFQAAAAALjaUXQAAAAAGIeiAwAAAMA4FB0AAAAAxqHoAAAAADAORQcAAACAcSg6AAAAAIxD0QEAAABgHIoOAAAAAONQdAAAAAAYh6IDAAAAwDgUHQAAAADGoegAAAAAMA5FBwAAAIBxKDoAAAAAjEPRAQAAAGAcig4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHEoOgAAAACMQ9EBAAAAYByKDgAAAADjUHQAAAAAGIeiAwAAAMA4FB0AAAAAxqHoAAAAADAORQcAAACAcSg6AAAAAIxD0QEAAABgHIoOAAAAAONQdAAAAAAYh6IDAAAAwDgUHQAAAADGoegAAAAAMA5FBwAAAIBxKDoAAAAAjEPRAQAAAGAcig4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHEoOgAAAACMQ9EBAAAAYByKDgAAAADjUHQAAAAAGIeiAwAAAMA4FB0AAAAAxrmiojNz5kwFBwerUqVK6tChgzZu3FjqvrNmzVKnTp1Uo0YN1ahRQ5GRkZfcHwAAAAD+qjIXnYULFyo6OloxMTHaunWrWrVqpe7du+vEiRMl7r9q1SoNGDBAK1eu1Pr16xUUFKTbbrtNx44d+8vhAQAAAKAkZS4606ZN07BhwxQVFaXQ0FDFx8fLx8dHc+fOLXH/Dz74QE888YRat26tpk2bavbs2bLb7UpMTPzL4QEAAACgJGUqOnl5edqyZYsiIyN/ewKbTZGRkVq/fv1lPcf58+f1yy+/qGbNmqXuk5ubq6ysrCI3AAAAALhcZSo6p06dUkFBgQIDA4uMBwYGKj09/bKeY8yYMapTp06RsvRHsbGx8vPzc9yCgoLKEhMAAACAm3PpqmtTp07VggUL9Mknn6hSpUql7jdu3DhlZmY6bmlpaS5MCQAAAOB/XYWy7Ozv7y9PT09lZGQUGc/IyFCtWrUu+djXXntNU6dO1TfffKOwsLBL7uvt7S1vb++yRAMAAAAAhzLN6Hh5ealdu3ZFFhL4dWGBiIiIUh/36quv6sUXX9SyZcsUHh5+5WkBAAAA4DKUaUZHkqKjozVkyBCFh4erffv2iouLU05OjqKioiRJgwcPVt26dRUbGytJeuWVVzRx4kR9+OGHCg4OdpzLU7VqVVWtWrUcvxQAAAAAuKjMRad///46efKkJk6cqPT0dLVu3VrLli1zLFCQmpoqm+23iaK33npLeXl5uueee4o8T0xMjF544YW/lh4AAAAASlDmoiNJI0aM0IgRI0rctmrVqiL3Dx8+fCUvAQAAAABXzKWrrgEAAACAK1B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHEoOgAAAACMQ9EBAAAAYByKDgAAAADjUHQAAAAAGIeiAwAAAMA4FB0AAAAAxqHoAAAAADAORQcAAACAcSg6AAAAAIxD0QEAAABgHIoOAAAAAONQdAAAAAAYh6IDAAAAwDgUHQAAAADGoegAAAAAMA5FBwAAAIBxKDoAAAAAjEPRAQAAAGAcig4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHEoOgAAAACMQ9EBAAAAYByKDgAAAADjUHQAAAAAGIeiAwAAAMA4FB0AAAAAxqHoAAAAADAORQcAAACAcSg6AAAAAIxD0QEAAABgHIoOAAAAAONQdAAAAAAYh6IDAAAAwDgUHQAAAADGoegAAAAAMA5FBwAAAIBxKDoAAAAAjEPRAQAAAGAcig4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMa5oqIzc+ZMBQcHq1KlSurQoYM2btxY6r47d+7U3XffreDgYHl4eCguLu5KswIAAADAZSlz0Vm4cKGio6MVExOjrVu3qlWrVurevbtOnDhR4v7nz59XSEiIpk6dqlq1av3lwAAAAADwZ8pcdKZNm6Zhw4YpKipKoaGhio+Pl4+Pj+bOnVvi/jfccIP++c9/6r777pO3t/dfDgwAAAAAf6ZMRScvL09btmxRZGTkb09gsykyMlLr168vt1C5ubnKysoqcgMAAACAy1WmonPq1CkVFBQoMDCwyHhgYKDS09PLLVRsbKz8/Pwct6CgoHJ7bgAAAADmuypXXRs3bpwyMzMdt7S0NKsjAQAAAPgfUqEsO/v7+8vT01MZGRlFxjMyMsp1oQFvb2/O5wEAAABwxco0o+Pl5aV27dopMTHRMWa325WYmKiIiIhyDwcAAAAAV6JMMzqSFB0drSFDhig8PFzt27dXXFyccnJyFBUVJUkaPHiw6tatq9jYWEkXFzBISUlx/PexY8eUlJSkqlWrqlGjRuX4pQAAAADARWUuOv3799fJkyc1ceJEpaenq3Xr1lq2bJljgYLU1FTZbL9NFP34449q06aN4/5rr72m1157TV26dNGqVav++lcAAAAAAH9Q5qIjSSNGjNCIESNK3PbH8hIcHKzCwsIreRkAAAAAuCJX5aprAAAAAPBXUHQAAAAAGIeiAwAAAMA4FB0AAAAAxqHoAAAAADAORQcAAACAcSg6AAAAAIxD0QEAAABgHIoOAAAAAONQdAAAAAAYh6IDAAAAwDgUHQAAAADGoegAAAAAMA5FBwAAAIBxKDoAAAAAjEPRAQAAAGAcig4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHEoOgAAAACMQ9EBAAAAYByKDgAAAADjUHQAAAAAGIeiAwAAAMA4FB0AAAAAxqHoAAAAADAORQcAAACAcSg6AAAAAIxD0QEAAABgHIoOAAAAAONQdAAAAAAYh6IDAAAAwDgUHQAAAADGoegAAAAAMA5FBwAAAIBxKDoAAAAAjEPRAQAAAGAcig4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHEoOgAAAACMQ9EBAAAAYByKDgAAAADjUHQAAAAAGIeiAwAAAMA4FB0AAAAAxrmiojNz5kwFBwerUqVK6tChgzZu3HjJ/RctWqSmTZuqUqVKatmypZYuXXpFYQEAAADgcpS56CxcuFDR0dGKiYnR1q1b1apVK3Xv3l0nTpwocf9169ZpwIABGjp0qH744Qf17dtXffv21Y4dO/5yeAAAAAAoSZmLzrRp0zRs2DBFRUUpNDRU8fHx8vHx0dy5c0vc/4033lCPHj00evRoNWvWTC+++KLatm2rN9988y+HBwAAAICSVCjLznl5edqyZYvGjRvnGLPZbIqMjNT69etLfMz69esVHR1dZKx79+769NNPS32d3Nxc5ebmOu5nZmZKkrKyssoSt0zsueed9tyu4MzvjavwHliP98B6vAfW+l///ku8B1cD3gPr8R5Yz5nvwa/PXVhYeMn9ylR0Tp06pYKCAgUGBhYZDwwM1O7du0t8THp6eon7p6enl/o6sbGxmjRpUrHxoKCgssR1K35xVicA74H1eA+sx3tgPd4D6/EeWI/3wHqueA/OnTsnPz+/UreXqei4yrhx44rMAtntdp05c0bXXHONPDw8LEx2ZbKyshQUFKS0tDRVq1bN6jhuiffAerwH1uM9sB7vgfV4D6zF9996JrwHhYWFOnfunOrUqXPJ/cpUdPz9/eXp6amMjIwi4xkZGapVq1aJj6lVq1aZ9pckb29veXt7FxmrXr16WaJelapVq/Y/+z+UKXgPrMd7YD3eA+vxHliP98BafP+t97/+HlxqJudXZVqMwMvLS+3atVNiYqJjzG63KzExURERESU+JiIiosj+krR8+fJS9wcAAACAv6rMh65FR0dryJAhCg8PV/v27RUXF6ecnBxFRUVJkgYPHqy6desqNjZWkjRq1Ch16dJFr7/+unr37q0FCxZo8+bNevvtt8v3KwEAAACA/1PmotO/f3+dPHlSEydOVHp6ulq3bq1ly5Y5FhxITU2VzfbbRNGNN96oDz/8UOPHj9dzzz2nxo0b69NPP1WLFi3K76u4ynl7eysmJqbY4XhwHd4D6/EeWI/3wHq8B9bjPbAW33/rudN74FH4Z+uyAQAAAMD/mDJfMBQAAAAArnYUHQAAAADGoegAAAAAMA5FBwAAAIBxKDoAAAAAjEPRKWe//PKLGjZsqF27dlkdBQB04MABjR8/XgMGDNCJEyckSV999ZV27txpcTIAAJyrzNfRwaVVrFhRFy5csDoGJG3evFkff/yxUlNTlZeXV2RbQkKCRakA11m9erV69uypm266Sd9++61efvllBQQEKDk5WXPmzNHixYutjgg4zeeff17iuJ+fn66//nrVrl3bxYkAuBozOk4wfPhwvfLKK8rPz7c6ittasGCBbrzxRu3atUuffPKJfvnlF+3cuVMrVqyQn5+f1fHcTl5envbs2cPPhIuNHTtWL730kpYvXy4vLy/HeLdu3bRhwwYLkwHO17dv3xJvN998s+rVq6dBgwbp/PnzVscEnC4nJ0cTJ05UixYtVLVqVfn6+iosLEyTJ082/meAC4Y6Qb9+/ZSYmKiqVauqZcuWqlKlSpHtzCY4X1hYmB599FENHz5cvr6+Sk5OVoMGDfToo4+qdu3amjRpktUR3cL58+c1cuRIvfvuu5KkvXv3KiQkRCNHjlTdunU1duxYixOarWrVqtq+fbsaNGjg+DkICQnR4cOH1bRpU2afneyhhx66rP3mzp3r5CT4vczMTG3ZskXDhw9Xv379NGXKFKsjGeuuu+667H35bOQceXl5uvHGG7Vjxw717NlTTZs2VWFhoXbt2qVly5apbdu2+vbbb1WxYkWrozoFh645QfXq1XX33XdbHcOtHThwQL1795YkeXl5KScnRx4eHnrqqafUrVs3io6LjBs3TsnJyVq1apV69OjhGI+MjNQLL7xA0XGy6tWr6/jx42rQoEGR8R9++EF169a1KJX7mDdvnurXr682bdqIvylePfz8/NStWzdNnz5d//jHPyg6TsQRFNZ76623dPToUSUnJ6tJkyZFtu3evVtdu3ZVfHy8Ro4caVFC56LoOME777xjdQS3V6NGDZ07d06SVLduXe3YsUMtW7bU2bNnjZ+mvZp8+umnWrhwoTp27CgPDw/HePPmzXXgwAELk7mH++67T2PGjNGiRYvk4eEhu92utWvX6plnntHgwYOtjme8xx9/XB999JEOHTqkqKgo3X///apZs6bVsfB/mjZtqqNHj1odw2h8HrJeQkKCJkyYUKzkSBd/Bp5//nktXrzY2KLDOTpOkp+fr2+++Ub//ve/HR+4f/zxR2VnZ1uczD107txZy5cvlyTde++9GjVqlIYNG6YBAwbolltusTid+zh58qQCAgKKjf86wwbnmjJlipo2baqgoCBlZ2crNDRUnTt31o033qjx48dbHc94M2fO1PHjx/Xss8/qiy++UFBQkP7+97/r66+/ZobnKnDw4EHVqVPH6hiAU6WkpKhr166lbr/55puVkpLiukAuxjk6TnDkyBH16NFDqampys3NdZyXMGrUKOXm5io+Pt7qiMY7c+aMLly4oDp16shut+vVV1/VunXr1LhxY40fP141atSwOqJb6Ny5s+69916NHDlSvr6+2rZtmxo0aKCRI0dq3759WrZsmdUR3UJaWpq2b9+u7OxstWnTRo0bN7Y6kls6cuSI5s2bp/nz5ys/P187d+5U1apVrY7llpKSkvTQQw+pS5cumj59utVx3MbixYtLXQ1169atFqUyW8WKFZWWlqZatWqVuP348eOqX79+sffDFBy65gSjRo1SeHi4kpOTdc011zjG+/Xrp2HDhlmYzH38/vAQm83GuSAWmTJlinr27KmUlBTl5+frjTfeUEpKitatW6fVq1dbHc9tBAUFKSgoSAUFBdq+fbt++uknyr4FbDabPDw8VFhYqIKCAqvjGK9GjRolzhzn5OQoPz9ft956K+drutCMGTP0/PPP68EHH9Rnn32mqKgoHThwQJs2bdLw4cOtjmcsu90uT0/PUrfbbDaj/z1iRscJrrnmGq1bt05NmjQpttJRaGgo54i4gKenp44fP17ssKnTp08rICDA6B/qq82BAwc0depUJScnKzs7W23bttWYMWPUsmVLq6MZ7x//+IdatmypoUOHqqCgQF26dNG6devk4+OjL7/88pKHM6B85ObmKiEhQXPnztWaNWt0++23KyoqSj169JDNxtHjzvTrao9/VK1aNTVp0kShoaEuTuTemjZtqpiYGA0YMKDIZ6OJEyfqzJkzevPNN62OaCSbzaYWLVqoQoWS5zZ+nV029XMRMzpOYLfbS/wf5ujRo/L19bUgkfsprb/n5uYWuZ4InK9hw4aaNWuW1THc0uLFi3X//fdLkr744gsdPHhQu3fv1nvvvafnn39ea9eutTih2Z544gktWLBAQUFBeuihh/TRRx/J39/f6lhuY8iQIVZHwO+kpqbqxhtvlCRVrlzZcf7yAw88oI4dO1J0nCQmJuZP9zF5pWCKjhPcdtttiouL09tvvy1J8vDwUHZ2tmJiYtSrVy+L05ltxowZki5+z2fPnl3k+PeCggJ9++23atq0qVXx3JLdbtf+/ft14sQJ2e32Its6d+5sUSr3cOrUKcdx2UuXLtXf//53XX/99XrooYf0xhtvWJzOfPHx8bruuusUEhKi1atXl3q4JtcPscbWrVs1ceJEffnll1ZHcQu1atXSmTNnVL9+fV133XXasGGDWrVqpUOHDrE4hxNdTtExGUXHCV5//XV1795doaGhunDhggYOHKh9+/bJ399fH330kdXxjPbrSaWFhYWKj48vclyql5eXgoODWQzChTZs2KCBAwfqyJEjxX6ReXh4GDtVfrUIDAxUSkqKateurWXLlumtt96SdPFCrpc6ZhvlY/DgwawuaLGvv/5ay5cvl5eXlx5++GGFhIRo9+7dGjt2rL744gt1797d6ohuo1u3bvr888/Vpk0bRUVF6amnntLixYu1efPmMl1YFCgLztFxkvz8fC1YsEDbtm1znJcwaNAgVa5c2epobuHmm29WQkICJ1xbrHXr1rr++us1adIk1a5du9iHPi4m51wvvPCC4uLiVLt2bZ0/f1579+6Vt7e35s6dq1mzZmn9+vVWRwScZs6cORo2bJhq1qypn376Sddcc42mTZumkSNHqn///ho1apSaNWtmdUy3YbfbZbfbHeeKLFiwwLEa6qOPPsph5RbZtWuXevfurYMHD1odxSkoOgCcpkqVKkpOTlajRo2sjuK2Fi9erLS0NN17772qV6+epIsnaVevXl133nmnxenMVtqiKHCNsLAwPfDAAxo9erT+85//6N5771XHjh318ccfO34WAHeXnJystm3bGnuEBUWnnHz++eeXve8dd9zhxCT41dGjR/X555+XuF7/tGnTLErlXrp166Znn31WPXr0sDoK4HI2m03p6ekUHYtUqVJFO3fuVHBwsAoLC+Xt7a2VK1fqpptusjqa2zp79qw2btxY4jmbgwcPtiiV2aKjoy+5/eTJk/rwww+NLTqco1NO+vbtW+T+r9dK+OOYJGP/Z7qaJCYm6o477nAcj92iRQsdPnxYhYWFatu2rdXx3MbIkSP19NNPKz09XS1btlTFihWLbA8LC7MomftITExUYmJiiR8s5s6da1EqwPl+/vln+fj4SLr4+9fb21u1a9e2OJX7+uKLLzRo0CBlZ2erWrVqRQ5l9vDwoOg4yRtvvKHWrVurWrVqJW7Pzs52cSLXouiUk99/gPjmm280ZswYTZkyRREREZKk9evXa/z48ZoyZYpVEd3KuHHj9Mwzz2jSpEny9fXVf/7zHwUEBGjQoEHMLrjQr0tWPvTQQ46xX/8IwGIEzjdp0iRNnjxZ4eHhJZ4jBef74+qPJXnyySddlMb9/P77n5+fr3nz5hVb4pvvv2s8/fTTeuihhzRlyhRHAYXzNWrUSE899ZTjUgN/lJSUpHbt2rk4letw6JoTtGjRQvHx8frb3/5WZPy7777TI488ol27dlmUzH34+voqKSlJDRs2VI0aNbRmzRo1b95cycnJuvPOO3X48GGrI7qFI0eOXHJ7/fr1XZTEPdWuXVuvvvqqHnjgAaujuCWbzaZ69epdcoU7Dw8PY08CtlpwcPCflnu+/65TpUoVbd++XSEhIVZHcSuDBg1SQECAY1XaP0pOTlabNm2KzfibghkdJzhw4ICqV69ebNzPz48P2C5SpUoVx3k5tWvX1oEDB9S8eXNJF68tAtegyFgrLy/PcYE+WGPz5s2co2MRft9eXbp3767NmzdTdFzs9ddfV25ubqnbW7VqZWzJkSg6TnHDDTcoOjpa7733ngIDAyVJGRkZGj16tNq3b29xOvfQsWNHrVmzRs2aNVOvXr309NNPa/v27UpISFDHjh2tjme0zz//XD179lTFihX/dJEOFuZwrocfflgffvihJkyYYHUUt8Shgta6cOGCvvnmG91+++2SLh7S/PsPfBUqVNDkyZNVqVIlqyK6ld69e2v06NFKSUkp8ZxNfh84x68XjXZXHLrmBPv371e/fv20d+9eBQUFSZLS0tLUuHFjffrppyy16wIHDx5Udna2wsLClJOTo6efftqxXv+0adOYaXCi3680ZbPZSt2Pc3Scb9SoUZo/f77CwsIUFhZW7IMFqw86F6uuWSs+Pl5LlizRF198IeniIc3Nmzd3XM9u9+7dGj169J+uSoXywe8Daz388MO6//771bVrV6ujuBRFx0kKCwu1fPly7d69W5LUrFkzRUZG8hc+AC5z8803l7rNw8NDK1ascGEa9zNp0iSNHj2aE68t0qlTJz377LPq06ePpItFJzk52XHo1Pvvv6+ZM2dy4Vy4hTvvvFNff/21rr32Wt133326//771apVK6tjOR1FB0YqLCzUli1bdPjwYXl4eKhBgwZq06YNRROAy5w6dUo5OTlFZpB37typ1157TTk5Oerbt68GDhxoYUKz1a5dW+vXr1dwcLAk6dprr9WmTZsc9/fu3asbbrhBmZmZ1oUEXOinn37SokWL9OGHH+q7775T06ZNNWjQIA0cONDxc2Eaio6TrF69Wq+99ppjhbXQ0FCNHj1anTp1sjiZ+VauXKmhQ4fqyJEjjmsZ/Vp25s6dq86dO1uc0GwzZsy47H1Z1tU19u/frwMHDqhz586qXLmyY3lvONeAAQNUp04dvf7665KkEydOqGnTpqpTp44aNmyor776SnPmzGFVPCepXLmykpKS1KRJkxK37969W61bt9aFCxdcnMx98dno6nH06FF99NFHmjt3rvbt26f8/HyrIzkFixE4wfvvv6+oqCjdddddjg9ya9as0S233KJ58+bxFzwn2r9/v26//XZ16NBB06dPV9OmTVVYWKiUlBTNmDFDvXr10rZt21j1xYlKW8Lyjzw8PCg6Tnb69Gn9/e9/18qVK+Xh4aF9+/YpJCREQ4cOVY0aNRwfwOEcGzZs0Lx58xz358+fr5o1ayopKUkVKlTQa6+9ppkzZ1J0nKRevXrasWNHqUVn27ZtqlevnotTua+SPhutXbuWz0YW+OWXX7R582Z9//33Onz4sGPhLCMVotw1bdq0cNq0acXGX3/99cKmTZtakMh9DB8+vLBbt24lbrPb7YXdunUrHDFihItTAdZ44IEHCrt3716YlpZWWLVq1cIDBw4UFhYWFi5btqwwNDTU4nTmq1SpUuHhw4cd93v27Fk4evRox/09e/YU1qxZ04pobuHJJ58sDA0NLfz555+LbTt//nxhaGho4ZNPPmlBMvfEZyPrrVixovDhhx8urFGjRqGfn19hVFRU4TfffFNot9utjuY0HLrmBN7e3tq5c2ex1dX279+vFi1aME3uRC1atFBsbKzj5NM/+uKLLzRu3Djt2LHDxcncW15eng4dOqSGDRuqQgUmkl2lVq1a+vrrr9WqVasiJ2IfPHhQYWFhys7Otjqi0QIDA/Xf//7XccKvv7+//v3vf+vuu++WJO3bt09t2rThfXCSjIwMtW7dWl5eXhoxYoSuv/56SdKePXv05ptvKj8/Xz/88IPZf82+ivDZyFp169bVmTNn1KNHDw0aNEh9+vSRt7e31bGcrvS1/nDFgoKClJiYWGz8m2++cSw3DedITU1Vy5YtS93eokULHTlyxIWJ3Nv58+c1dOhQ+fj4qHnz5kpNTZUkjRw5UlOnTrU4nflycnJKXPHrzJkzbvELzmodO3bUjBkzZLfbtXjxYp07d07dunVzbP/9JQhQ/gIDA7Vu3To1a9ZMY8eOVb9+/dSvXz+NGzdOoaGhWrNmDSXHhfhsZK0XXnhBx48f1yeffKJ77rnHbX4H8KdVJ3j66af15JNPKikpyXFV8rVr12revHl64403LE5ntuzs7Esu5erj46Pz58+7MJF7GzdunJKTk7Vq1Sr16NHDMR4ZGakXXnhBY8eOtTCd+Tp16qT58+frxRdflHTxvCi73a5XX331kktPo3y8+OKLuuWWW/T+++8rPz9fzz33nGrUqOHYvmDBAnXp0sXChOZr0KCBli1bpjNnzmj//v2SpEaNGqlmzZoWJ3M/fDay1rBhw6yOYAkOXXOSTz75RK+//rpjZZFmzZpp9OjRuvPOOy1OZjabzaYVK1aU+kvs1KlTuvXWW7kwmYvUr19fCxcuVMeOHYscOrV//361bdtWWVlZVkc02o4dO3TLLbeobdu2WrFihe644w7t3LlTZ86c0dq1a9WwYUOrIxrv1KlTWrt2rWrVqqUOHToU2fb+++9r3rx5+uabbyxKB7gWn42sk5OTo6lTpyoxMVEnTpyQ3W4vsv3gwYMWJXMuig6MYrPZ5OHhoZL+t/51nCswu46Pj4927NihkJCQIkUnOTlZnTt35voVLpCZmak333xTycnJys7OVtu2bTV8+HDVrl3b6mhuLzk5WW3btuXfIwBON2DAAK1evVoPPPCAateuXewSA6NGjbIomXNx6JoTbNq0SXa7vdhf777//nt5enoqPDzcomTmO3TokNUR8Dvh4eFasmSJRo4cKUmOf1hnz56tiIgIK6O5DT8/Pz3//PNWxwAAWOirr77SkiVLdNNNN1kdxaUoOk4wfPhwPfvss8WKzrFjx/TKK6/o+++/tyiZ+X5/BXJYb8qUKerZs6dSUlKUn5+vN954QykpKVq3bp1Wr15tdTzj7du3T5999pkOHz4sDw8PhYSEqG/fvmrQoIHV0QC4gZo1a2rv3r3y9/dXjRo1Lnmh4jNnzrgwmfupUaOGW56bxqFrTlC1atUSL0p56NAhhYWF6dy5cxYlA1zvwIEDmjp1apFDp8aMGXPJ1fHw18XGxmrixImy2+0KCAhQYWGhTp48KU9PT02ZMkXPPPOM1RHdHoeuwXTvvvuu7rvvPnl7e2vevHmXLDpDhgxxYTL38/777+uzzz7Tu+++e8lFm0xD0XGCa665Rl9++WWxQ3PWrVun3r1766effrIoGQB3sHLlSkVGRmrChAkaNWqUY6WvM2fOKC4uTlOmTNGKFSvUuXNni5Oa7a677rrk9rNnz2r16tUUHQBO16ZNGx04cECFhYUKDg5WxYoVi2zfunWrRcmci6LjBAMGDNDx48f12Wefyc/PT9LFX2h9+/ZVQECAPv74Y4sTAq6xdetWVaxY0TF789lnn+mdd95RaGioXnjhBXl5eVmc0Ez9+/dX9erV9e9//7vE7Y888ojOnTunjz76yMXJ3EtUVNRl7ffOO+84OQlgPU9PTx0/flwBAQFFxk+fPq2AgAAKv5NNmjTpkttjYmJclMS1KDpOcOzYMXXu3FmnT59WmzZtJElJSUkKDAzU8uXLuTAW3MYNN9ygsWPH6u6779bBgwcVGhqqu+66S5s2bVLv3r0VFxdndUQjNWjQQO+9957+9re/lbj9u+++0+DBg1m8A4DL2Gw2paenFys6P/74oxo2bKiff/7ZomQwGYsROEHdunW1bds2ffDBB0pOTlblypUVFRWlAQMGFJsqhHN069ZNCQkJql69epHxrKws9e3bVytWrLAmmJvZu3evWrduLUlatGiRunTpog8//FBr167VfffdR9FxkoyMDAUHB5e6vUGDBkpPT3ddIABua8aMGZIurro5e/ZsVa1a1bGtoKBA3377rZo2bWpVPLezZcsWx3WMmjdv7viDvKkoOk5SpUoVPfLII1bHcFurVq1SXl5esfELFy7ou+++syCReyosLHRclOybb77R7bffLkkKCgrSqVOnrIxmtAsXLlzysMCKFSuW+PMBAOVt+vTpki7+PoiPj5enp6djm5eXl4KDgxUfH29VPLdx4sQJ3XfffVq1apXjj8Bnz57VzTffrAULFujaa6+1NqCTUHScZN++fVq5cmWJV5+dOHGiRanMt23bNsd/p6SkFPmrdUFBgZYtW6a6detaEc0thYeH66WXXlJkZKRWr16tt956S9LFFQgDAwMtTme2P/7l9PdY+RGAq/x6iOzNN9+shIQEx+IocK2RI0fq3Llz2rlzp5o1aybp4uekIUOG6MknnzT2nE3O0XGCWbNm6fHHH5e/v79q1apVZDlFDw8PY1e2uBrYbDbH97uk/7UrV66sf/3rX3rooYdcHc0tbdu2TYMGDVJqaqqio6MdJzuOHDlSp0+f1ocffmhxQjMFBwdfchnXX3GODgC4Bz8/P33zzTe64YYbioxv3LhRt912m86ePWtNMCej6DhB/fr19cQTT2jMmDFWR3E7R44cUWFhoUJCQrRx48YiU7FeXl4KCAgoMm0Oa1y4cEGenp6cswYAbuLuu+9W+/bti302evXVV7Vp0yYtWrTIomTuwdfXV999953jvNlf/fDDD+rSpYuysrKsCeZkFB0nqFatmpKSkopdMBQAAMAdXXvttVqxYkWxi0Vv375dkZGRysjIsCiZe7jzzjt19uxZffTRR6pTp46ki6sEDxo0SDVq1NAnn3xicULn4BwdJ7j33nv13//+V4899pjVUdwa50lZr6CgQNOnT9fHH3+s1NTUYifAnzlzxqJkAABXys7OLnGRlIoVKxo7m3A1efPNN3XHHXcoODjYcZmTtLQ0tWjRQu+//77F6ZyHouMEjRo10oQJE7Rhwwa1bNmy2OE5Tz75pEXJ3MefnSdF0XGNSZMmafbs2Xr66ac1fvx4Pf/88zp8+LA+/fRT3gMAcCMtW7bUwoULi/3bv2DBAoWGhlqUyn0EBQVp69at+uabb7R7925JUrNmzRQZGWlxMufi0DUnaNCgQanbPDw8dPDgQRemcU+cJ3V1aNiwoWbMmKHevXvL19dXSUlJjrENGzawGAEAuIkvvvhCd911lwYOHKhu3bpJkhITE/XRRx9p0aJF6tu3r7UBDbVixQqNGDFCGzZsULVq1Ypsy8zM1I033qj4+Hh16tTJooTORdGBkThP6upQpUoV7dq1S9ddd51q166tJUuWqG3btjp48KDatGmjzMxMqyMCAFxkyZIlmjJlipKSklS5cmWFhYUpJiZGXbp0sTqase644w7dfPPNeuqpp0rcPmPGDK1cudLYc3RsVgcAnOHX86RgrXr16un48eOSLs7u/PqebNq0Sd7e3lZGcxsHDhzQ+PHjNWDAAJ04cUKS9NVXX2nnzp0WJwPgbnr37q21a9cqJydHp06d0ooVKyg5TpacnKwePXqUuv22227Tli1bXJjItSg65Sg0NLTIydVPPPFEkau/nzhxQj4+PlZEczu/nif14IMP6vXXX9eMGTOK3OAa/fr1U2JioqSL186ZMGGCGjdurMGDB3MtIxdYvXq1WrZsqe+//14JCQnKzs6WdPEX36/XNAIAVzl79qxmz56t5557zvF5aevWrTp27JjFycyVkZFxyUs5VKhQQSdPnnRhItfi0LVyZLPZlJ6eroCAAEnFD5/KyMhQ7dq1i60AhvLHeVJXpw0bNmjdunVq3Lix+vTpY3Uc40VEROjee+9VdHS0fH19lZyc7LjG1F133aWjR49aHRGAm9i2bZsiIyPl5+enw4cPa8+ePQoJCdH48eOVmpqq+fPnWx3RSA0bNtTrr79e6jlQCQkJeuaZZ4z9XMSMjhOV1CEv52rl+OsOHTpU6s3UH+ar0enTpx3/nZaWpqVLl+r48ePy8/OzMJX72L59u/r161dsPCAgoMhsMwA4W3R0tB588EHt27dPlSpVcoz36tVL3377rYXJzNarVy9NmDBBFy5cKLbt559/VkxMjG6//XYLkrkGRQdGy8vL0549e5Sfn291FLeyfft2BQcHKyAgQE2bNlVSUpJuuOEGTZ8+XW+//ba6deumTz/91OqYxqtevbrjHKnf++GHH1S3bl0LEgFwV5s2bdKjjz5abLxu3bpKT0+3IJF7GD9+vM6cOaPrr79er776qj777DN99tlneuWVV9SkSROdOXNGzz//vNUxnYaiU448PDyKzdgwg2ON8+fPa+jQofLx8VHz5s2Vmpoq6eJ5IlOnTrU4nfmeffZZtWzZUt9++626du2q22+/Xb1791ZmZqZ++uknPfroo7wPLnDfffdpzJgxSk9Pl4eHh+x2u9auXatnnnlGgwcPtjoeADfi7e1d4oVB9+7dq2uvvdaCRO4hMDBQ69atU4sWLTRu3Dj169dP/fr103PPPacWLVpozZo1CgwMtDqm03COTjmy2Wxq0aKFKlS4eB3Wbdu2qWnTpo4rAefn52vnzp0qKCiwMqZbGDVqlNauXau4uDj16NFD27ZtU0hIiD777DO98MIL+uGHH6yOaDR/f3+tWLFCYWFhys7OVrVq1bRp0ya1a9dOkrR792517NhRZ8+etTao4fLy8jR8+HDNmzdPBQUFqlChggoKCjRw4EDNmzdPnp6eVkcE4CYefvhhnT59Wh9//LFq1qypbdu2ydPTU3379lXnzp0VFxdndUTj/fTTT9q/f78KCwvVuHFj1ahRw+pITkfRKUeTJk26rP1Y7cj56tevr4ULF6pjx45FTsLev3+/2rZtW+JflVB+/rgwx+/fA+niwhx16tSh9LtIamqqduzYoezsbLVp00aNGze2OhIAN5OZmal77rlHmzdv1rlz51SnTh2lp6crIiJCS5cuVZUqVayOCANVsDqASSgwV4+TJ086PmT/Xk5ODocTugiHcV49rrvuOl133XVWxwDgxvz8/LR8+XKtXbtWycnJys7OVtu2bRUZGWl1NBiMogMjhYeHa8mSJRo5cqSk3z5kz549WxEREVZGcxsPPvig46KgFy5c0GOPPeb4i11ubq6V0YwWHR192ftOmzbNiUkA4KJffvlFlStXVlJSkm666SbddNNNVkeCm6DowEhTpkxRz549lZKSovz8fL3xxhtKSUnRunXrtHr1aqvjGW/IkCFF7t9///3F9uFkeOe43PPPmGED4CoVK1bUddddx+HKcDnO0YGxDhw4oKlTpxaZIh8zZoxatmxpdTQAANzKnDlzlJCQoPfee081a9a0Og7cBEUHANxAWlqaJCkoKMjiJADcUZs2bbR//3798ssvql+/frHFB7Zu3WpRMpiMQ9ec7MKFC0WuAAzXWLp0qTw9PdW9e/ci419//bXsdrt69uxpUTLAdfLz8zVp0iTNmDFD2dnZkqSqVatq5MiRiomJUcWKFS1OCMBd9O3b1+oIcEPM6DiB3W7Xyy+/rPj4eGVkZGjv3r0KCQnRhAkTFBwcrKFDh1od0XhhYWGaOnWqevXqVWR82bJlGjNmjJKTky1KBrjO448/roSEBE2ePNmxCMf69ev1wgsvqG/fvnrrrbcsTggAgPNQdJxg8uTJevfddzV58mQNGzZMO3bsUEhIiBYuXKi4uDitX7/e6ojGq1y5snbt2qXg4OAi44cPH1bz5s2Vk5NjTTDAhfz8/LRgwYJiM5hLly7VgAEDlJmZaVEyAACcz2Z1ABPNnz9fb7/9tgYNGlTkyuOtWrXS7t27LUzmPvz8/HTw4MFi4/v37+eiZHAb3t7excq+JDVo0EBeXl6uDwTArdSsWVOnTp2SJNWoUUM1a9Ys9QY4A+foOMGxY8fUqFGjYuN2u12//PKLBYncz5133ql//OMf+uSTT9SwYUNJF0vO008/rTvuuMPidIBrjBgxQi+++KLeeecdxzWNcnNz9fLLL2vEiBEWpwNguunTp8vX11eSFBcXZ20YuCUOXXOCdu3a6amnntL9998vX19fJScnKyQkRJMnT9by5cv13XffWR3ReJmZmerRo4c2b96sevXqSZKOHj2qTp06KSEhQdWrV7c2IOAC/fr1U2Jiory9vdWqVStJUnJysvLy8nTLLbcU2TchIcGKiAAAOA0zOk4wceJEDRkyRMeOHZPdbldCQoL27Nmj+fPn68svv7Q6nlvw8/PTunXrtHz5ciUnJ6ty5coKCwtT586drY4GuEz16tV19913FxljeWkArpKVlXXZ+1arVs2JSeCumNFxku+++06TJ08ucrHKiRMn6rbbbrM6mvF++eUXVa5cWUlJSWrRooXVcQAAcEs2m00eHh6XtW9BQYGT08AdMaPjJJ06ddLy5cutjuGWKlasqOuuu45/NAEAsNDKlSsd/3348GGNHTtWDz74YJHl7t99913FxsZaFRGGY0bHCTZt2iS73a4OHToUGf/+++/l6emp8PBwi5K5jzlz5ighIUHvvfceq7nAbZ0+fVoTJ07UypUrdeLECdnt9iLbz5w5Y1EyAO7mlltu0cMPP6wBAwYUGf/www/19ttva9WqVdYEg9EoOk7Qvn17Pfvss7rnnnuKjCckJOiVV17R999/b1Ey99GmTRvt379fv/zyi+rXr19sSemtW7dalAxwnV69emn//v0aOnSoAgMDix1CMmTIEIuSAXA3Pj4+Sk5OVuPGjYuM7927V61bt9b58+ctSgaTceiaE6SkpKht27bFxtu0aaOUlBQLErmfvn37Wh0BsNx3332nNWvWOFZcAwCrBAUFadasWXr11VeLjM+ePZtFUuA0FB0n8Pb2VkZGhkJCQoqMHz9+XBUq8C13hZiYGKsjAJZr2rSpfv75Z6tjAICmT5+uu+++W1999ZXj0P6NGzdq3759+s9//mNxOpiKQ9ecYMCAATp+/Lg+++wz+fn5SZLOnj2rvn37KiAgQB9//LHFCd3D2bNntXjxYh04cECjR49WzZo1tXXrVgUGBqpu3bpWxwOcbtOmTRo7dqwmTpyoFi1aqGLFikW2s5wrAFdKS0vTW2+9pd27d0uSmjVrpscee4wZHTgNRccJjh07ps6dO+v06dNq06aNJCkpKUmBgYFavnw5P9AusG3bNkVGRsrPz0+HDx/Wnj17FBISovHjxys1NVXz58+3OiLgdPv27dPAgQOLnZNWWFgoDw8PViYEABiNouMkOTk5+uCDD4pcrHLAgAHF/qIK54iMjFTbtm316quvytfXV8nJyQoJCdG6des0cOBAHT582OqIgNO1b99eFSpU0KhRo0pcjKBLly4WJQPgDrZt26YWLVrIZrNp27Ztl9w3LCzMRangTig6MJKfn5+2bt2qhg0bFik6R44cUZMmTXThwgWrIwJO5+Pjox9++EFNmjSxOgoAN2Sz2ZSenq6AgADHxUNL+tjJDDOchTPjnWTfvn2lXrti4sSJFqVyH97e3srKyio2vnfvXl177bUWJAJcLzw8XGlpaRQdAJY4dOiQ43fuoUOHLE4Dd8SMjhPMmjVLjz/+uPz9/VWrVq0ih4t4eHhwDRcXePjhh3X69Gl9/PHHqlmzprZt2yZPT0/17dtXnTt3VlxcnNURAadbtGiRXnjhBY0ePVotW7Ysdugsh4oAcJXTp0/rmmuukXRxUYJZs2bp559/1h133KFOnTpZnA6moug4Qf369fXEE09ozJgxVkdxW5mZmbrnnnu0efNmnTt3TnXq1FF6eroiIiK0dOnSYhcQBUxks9mKjf166AiHigBwhe3bt6tPnz5KS0tT48aNtWDBAvXo0UM5OTmy2WzKycnR4sWLuf4dnIKi4wTVqlVTUlJSsevowPXWrl2r5ORkZWdnq23btoqMjLQ6EuAyR44cueT2+vXruygJAHfVs2dPVahQQWPHjtV7772nL7/8Ut27d9esWbMkSSNHjtSWLVu0YcMGi5PCRBQdJxg6dKhuuOEGPfbYY1ZHcUsLFy7U559/rry8PN1yyy28DwAAWMTf318rVqxQWFiYsrOzVa1aNW3atEnt2rWTJO3evVsdO3bU2bNnrQ0KI7EYgRM0atRIEyZM0IYNG0o8Lv7JJ5+0KJn53nrrLQ0fPlyNGzdW5cqVlZCQoAMHDuif//yn1dEAl/j888/Vs2dPVaxYUZ9//vkl973jjjtclAqAuzpz5oxq1aolSapataqqVKmiGjVqOLbXqFFD586dsyoeDMeMjhM0aNCg1G0eHh46ePCgC9O4l+bNm+vvf/+7YmJiJEnvv/++Hn30UeXk5FicDHCNPy7nWhrO0QHgCjabTRkZGY7V13x9fbVt2zbHZ6WMjAzVqVOHf4/gFBQdGKVy5cratWuXgoODJUl2u12VK1fW4cOHVbt2bWvDAQDgZmw2m3r27Clvb29J0hdffKFu3bo5FgXKzc3VsmXLKDpwCg5dc6K8vDwdOnRIDRs2VIUKfKtdITc3t8iKajabTV5eXvr5558tTAUAgHsaMmRIkfv3339/sX0GDx7sqjhwM8zoOMH58+c1cuRIvfvuu5IuXqQyJCREI0eOVN26dTV27FiLE5rLZrPpkUcekY+Pj2Ns5syZuv/+++Xn5+cYmzZtmhXxAJdYv369Tp8+rdtvv90xNn/+fMXExCgnJ0d9+/bVv/71L8dfWAEAMFHpB3Djio0bN07JyclatWqVKlWq5BiPjIzUwoULLUxmvs6dO2vPnj364YcfHLcbb7xRBw8edNxPSkqyOibgVJMnT9bOnTsd97dv366hQ4cqMjJSY8eO1RdffKHY2FgLEwIA4HzM6DhB/fr1tXDhQnXs2FG+vr5KTk5WSEiI9u/fr7Zt2yorK8vqiAAMVrt2bX3xxRcKDw+XJD3//PNavXq11qxZI0latGiRYmJilJKSYmVMAACcihkdJzh58qQCAgKKjefk5MjDw8OCRADcyU8//aTAwEDH/dWrV6tnz56O+zfccIPS0tKsiAYAgMtQdJwgPDxcS5Yscdz/tdzMnj1bERERVsUC4CYCAwN16NAhSRcXRdm6das6duzo2H7u3Lli1/cCAMA0LAXmBFOmTFHPnj2VkpKi/Px8vfHGG0pJSdG6deu0evVqq+MBMFyvXr00duxYvfLKK/r000/l4+OjTp06ObZv27ZNDRs2tDAhAADOx4yOE/ztb39TUlKS8vPz1bJlS/33v/9VQECA1q9fr3bt2lkdD4DhXnzxRVWoUEFdunTRrFmzNGvWLHl5eTm2z507V7fddpuFCQEAcD4WIwAAQ2VmZqpq1ary9PQsMn7mzBlVrVq1SPkBAMA0FB0n8PT01PHjx4stSHD69GkFBARw9V8X2Lhxo9avX6/09HRJUq1atRQREaH27dtbnAwAAACuwDk6TlBad8zNzeUvqE524sQJ3X333Vq7dq2uu+46x8pTGRkZeuqpp3TTTTfpP//5T4mr4gEAAMAcFJ1yNGPGDEkXV1mbPXu2qlat6thWUFCgb7/9Vk2bNrUqnlt44oknVFBQoF27dqlJkyZFtu3Zs0cPPfSQhg8frkWLFlmUEAAAAK7AoWvlqEGDBpKkI0eOqF69ekWOi/fy8lJwcLAmT56sDh06WBXReL6+vvr222/Vpk2bErdv2bJFXbt21blz51ycDAAAAK7EjE45+vW6FTfffLMSEhJUo0YNixO5H29vb2VlZZW6/dy5c/L29nZhIgAAAFiB5aWdYOXKlZQci/Tv319DhgzRJ598UqTwZGVl6ZNPPlFUVJQGDBhgYUIAAAC4AjM6TlBQUKB58+YpMTFRJ06ckN1uL7J9xYoVFiUz37Rp02S323XfffcpPz/fsfhDXl6eKlSooKFDh+q1116zOCUAAACcjXN0nGDEiBGaN2+eevfurdq1a8vDw6PI9unTp1uUzH1kZWVpy5YtRZaXbteunapVq2ZxMgAAALgCRccJ/P39NX/+fPXq1cvqKAAAAIBb4hwdJ/Dy8lKjRo2sjoESZGRkaPLkyVbHAAAAgJMxo+MEr7/+ug4ePKg333yz2GFrsFZycrLatm2rgoICq6MAAADAiViMwAnWrFmjlStX6quvvlLz5s1VsWLFItsTEhIsSma+bdu2XXL7nj17XJQEAAAAVqLoOEH16tXVr18/q2O4pdatW8vDw0MlTVT+Os4sGwAAgPk4dA1G8ff316uvvqpbbrmlxO07d+5Unz59OHQNAADAcMzoOEl+fr5WrVqlAwcOaODAgfL19dWPP/6oatWqqWrVqlbHM1a7du30448/qn79+iVuP3v2bImzPQAAADALRccJjhw5oh49eig1NVW5ubm69dZb5evrq1deeUW5ubmKj4+3OqKxHnvsMeXk5JS6/brrrtM777zjwkQAAACwAoeuOUHfvn3l6+urOXPm6JprrlFycrJCQkK0atUqDRs2TPv27bM6IgAAAGA0ZnSc4LvvvtO6devk5eVVZDw4OFjHjh2zKBUAAADgPrhgqBPY7fYST3Y/evSofH19LUgEAAAAuBeKjhPcdtttiouLc9z38PBQdna2YmJi1KtXL+uCAQAAAG6Cc3Sc4OjRo+revbsKCwu1b98+hYeHa9++ffL399e3336rgIAAqyMCAAAARqPoOEl+fr4WLlyo5ORkZWdnq23btho0aJAqV65sdTQAAADAeBQdGGXw4MGaOXOm41yo5ORkhYaGqmLFihYnAwAAgCtxjk452rt3rzZu3FhkLDExUTfffLPat2+vKVOmWJTMfXzwwQf6+eefHfc7deqktLQ0CxMBAADAChSdcjRmzBh9+eWXjvuHDh1Snz595OXlpYiICMXGxhZZpADl748TlExYAgAAuCeuo1OONm/erGeffdZx/4MPPtD111+vr7/+WpIUFhamf/3rX/rHP/5hUUIAAADAPVB0ytGpU6dUr149x/2VK1eqT58+jvtdu3bV008/bUU0t5KSkqL09HRJF2d0du/erezs7CL7hIWFWRENAAAALkLRKUc1a9bU8ePHFRQUJLvdrs2bNys6OtqxPS8vj0OpXOCWW24p8n2+/fbbJV28nlFhYaE8PDxKvKArAAAAzEHRKUddu3bViy++qP/3//6fFi1aJLvdrq5duzq2p6SkKDg42LJ87uDQoUNWRwAAAMBVgKJTjl5++WXdeuutql+/vjw9PTVjxgxVqVLFsf29995Tt27dLExovnfffVfPPPOMfHx8rI4CAAAAC3EdnXKWn5+vnTt36tprr1WdOnWKbEtOTla9evV0zTXXWJTOfJ6enjp+/LgCAgKsjgIAAAALUXRgFJvNpvT0dIoOAACAm+M6OjCOh4eH1REAAABgMWZ0YBSbzSY/P78/LTtnzpxxUSIAAABYgcUIYJxJkybJz8/P6hgAAACwEDM6MArn6AAAAEDiHJ1yNXjwYJ07d85xPzk5Wb/88ouFidwP5+cAAABAouiUqw8++EA///yz436nTp2UlpZmYSL3wwQlAAAAJM7RKVd//JDNh27Xs9vtVkcAAADAVYAZHQAAAADGYUannKWkpCg9PV3SxRmd3bt3Kzs7u8g+YWFhVkQDAAAA3AarrpUjm80mDw+PEg9Z+3Xcw8NDBQUFFqQDAAAA3AczOuXo0KFDVkcAAAAAIIpOuXr33Xf1zDPPyMfHx+ooAAAAgFvj0LVy5OnpqePHj3OxSgAAAMBirLpWjuiMAAAAwNWBolPOPDw8rI4AAAAAuD0OXStHNptNfn5+f1p2zpw546JEAAAAgHtiMYJyNmnSJPn5+VkdAwAAAHBrzOiUI5vNpvT0dBYjAAAAACzGOTrliPNzAAAAgKsDRaccMTkGAAAAXB04dA0AAACAcZjRAQAAAGAcig4AAAAA41B0AAAAABiHogMAAADAOBQdAAAAAMah6AAAAAAwDkUHAAAAgHH+P6tOwqZAAJ3dAAAAAElFTkSuQmCC"},"metadata":{}}]},{"cell_type":"markdown","source":"## Model Ensembling ","metadata":{}},{"cell_type":"code","source":"baseline_pred_probs = np.max(model_0.predict_proba(val_sentences), axis = 1)\n\n#combining prediction probabilities\ncombined_pred_probs = baseline_pred_probs + tf.squeeze(model_2_pred_probs, axis=1) + tf.squeeze(model_6_pred_probs)\n\n#averaging the prediction\ncombined_preds = tf.round(combined_pred_probs/3)\ncombined_preds[:20]                       ","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:23.051707Z","iopub.execute_input":"2023-10-04T08:18:23.052056Z","iopub.status.idle":"2023-10-04T08:18:23.081462Z","shell.execute_reply.started":"2023-10-04T08:18:23.052029Z","shell.execute_reply":"2023-10-04T08:18:23.080322Z"},"trusted":true},"execution_count":202,"outputs":[{"execution_count":202,"output_type":"execute_result","data":{"text/plain":"<tf.Tensor: shape=(20,), dtype=float32, numpy=\narray([0., 1., 1., 0., 0., 1., 1., 1., 1., 0., 0., 0., 0., 1., 0., 0., 0.,\n       0., 0., 1.], dtype=float32)>"},"metadata":{}}]},{"cell_type":"code","source":"ensemble_results = calculate_results(val_labels, combined_preds)\nensemble_results","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:23.082845Z","iopub.execute_input":"2023-10-04T08:18:23.083179Z","iopub.status.idle":"2023-10-04T08:18:23.097552Z","shell.execute_reply.started":"2023-10-04T08:18:23.083109Z","shell.execute_reply":"2023-10-04T08:18:23.096780Z"},"trusted":true},"execution_count":203,"outputs":[{"execution_count":203,"output_type":"execute_result","data":{"text/plain":"{'accurracy': 79.00262467191601,\n 'precision': 0.7899413404218827,\n 'recall': 0.7900262467191601,\n 'f1': 0.7893166827702129}"},"metadata":{}}]},{"cell_type":"code","source":"ensemble_results[\"accurracy\"] = ensemble_results[\"accurracy\"] / 100\nensemble_results[\"accurracy\"] ","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:23.098522Z","iopub.execute_input":"2023-10-04T08:18:23.099460Z","iopub.status.idle":"2023-10-04T08:18:23.110620Z","shell.execute_reply.started":"2023-10-04T08:18:23.099428Z","shell.execute_reply":"2023-10-04T08:18:23.109595Z"},"trusted":true},"execution_count":204,"outputs":[{"execution_count":204,"output_type":"execute_result","data":{"text/plain":"0.7900262467191601"},"metadata":{}}]},{"cell_type":"code","source":"all_model_results.loc[\"Ensemble Results\"] = ensemble_results","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:23.111859Z","iopub.execute_input":"2023-10-04T08:18:23.112169Z","iopub.status.idle":"2023-10-04T08:18:23.127397Z","shell.execute_reply.started":"2023-10-04T08:18:23.112141Z","shell.execute_reply":"2023-10-04T08:18:23.126220Z"},"trusted":true},"execution_count":205,"outputs":[]},{"cell_type":"code","source":"all_model_results","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:23.128694Z","iopub.execute_input":"2023-10-04T08:18:23.129116Z","iopub.status.idle":"2023-10-04T08:18:23.149067Z","shell.execute_reply.started":"2023-10-04T08:18:23.129068Z","shell.execute_reply":"2023-10-04T08:18:23.148131Z"},"trusted":true},"execution_count":206,"outputs":[{"execution_count":206,"output_type":"execute_result","data":{"text/plain":"                     accurracy  precision    recall        f1\nBaseline              0.792651   0.811139  0.792651  0.786219\nSimple Dense          0.788714   0.793057  0.788714  0.785956\nLSTM                  0.772966   0.775649  0.772966  0.770438\nGRU                   0.770341   0.770288  0.770341  0.769333\nBidirectional         0.763780   0.763520  0.763780  0.763602\nConv1D                0.755906   0.755582  0.755906  0.754985\nTF Sentence Encoder   0.811024   0.820114  0.811024  0.810554\nTF 10 Percent Data    0.796588   0.804169  0.796588  0.793067\nEnsemble Results      0.790026   0.789941  0.790026  0.789317","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>accurracy</th>\n      <th>precision</th>\n      <th>recall</th>\n      <th>f1</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>Baseline</th>\n      <td>0.792651</td>\n      <td>0.811139</td>\n      <td>0.792651</td>\n      <td>0.786219</td>\n    </tr>\n    <tr>\n      <th>Simple Dense</th>\n      <td>0.788714</td>\n      <td>0.793057</td>\n      <td>0.788714</td>\n      <td>0.785956</td>\n    </tr>\n    <tr>\n      <th>LSTM</th>\n      <td>0.772966</td>\n      <td>0.775649</td>\n      <td>0.772966</td>\n      <td>0.770438</td>\n    </tr>\n    <tr>\n      <th>GRU</th>\n      <td>0.770341</td>\n      <td>0.770288</td>\n      <td>0.770341</td>\n      <td>0.769333</td>\n    </tr>\n    <tr>\n      <th>Bidirectional</th>\n      <td>0.763780</td>\n      <td>0.763520</td>\n      <td>0.763780</td>\n      <td>0.763602</td>\n    </tr>\n    <tr>\n      <th>Conv1D</th>\n      <td>0.755906</td>\n      <td>0.755582</td>\n      <td>0.755906</td>\n      <td>0.754985</td>\n    </tr>\n    <tr>\n      <th>TF Sentence Encoder</th>\n      <td>0.811024</td>\n      <td>0.820114</td>\n      <td>0.811024</td>\n      <td>0.810554</td>\n    </tr>\n    <tr>\n      <th>TF 10 Percent Data</th>\n      <td>0.796588</td>\n      <td>0.804169</td>\n      <td>0.796588</td>\n      <td>0.793067</td>\n    </tr>\n    <tr>\n      <th>Ensemble Results</th>\n      <td>0.790026</td>\n      <td>0.789941</td>\n      <td>0.790026</td>\n      <td>0.789317</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"val_df = pd.DataFrame({\"text\": val_sentences,\n                       \"target\": val_labels,\n                       \"pred\": model_7_preds,\n                       \"pred_prob\": tf.squeeze(model_7_pred_probs)})\nval_df.head()","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:23.150875Z","iopub.execute_input":"2023-10-04T08:18:23.151305Z","iopub.status.idle":"2023-10-04T08:18:23.172243Z","shell.execute_reply.started":"2023-10-04T08:18:23.151264Z","shell.execute_reply":"2023-10-04T08:18:23.171006Z"},"trusted":true},"execution_count":207,"outputs":[{"execution_count":207,"output_type":"execute_result","data":{"text/plain":"                                                text  target  pred  pred_prob\n0  DFR EP016 Monthly Meltdown - On Dnbheaven 2015...       0   0.0   0.268827\n1  FedEx no longer to transport bioterror germs i...       0   1.0   0.770848\n2  Gunmen kill four in El Salvador bus attack: Su...       1   1.0   1.035484\n3  @camilacabello97 Internally and externally scr...       1   0.0   0.116381\n4  Radiation emergency #preparedness starts with ...       1   1.0   0.623125","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>text</th>\n      <th>target</th>\n      <th>pred</th>\n      <th>pred_prob</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>DFR EP016 Monthly Meltdown - On Dnbheaven 2015...</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.268827</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>FedEx no longer to transport bioterror germs i...</td>\n      <td>0</td>\n      <td>1.0</td>\n      <td>0.770848</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>Gunmen kill four in El Salvador bus attack: Su...</td>\n      <td>1</td>\n      <td>1.0</td>\n      <td>1.035484</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>@camilacabello97 Internally and externally scr...</td>\n      <td>1</td>\n      <td>0.0</td>\n      <td>0.116381</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>Radiation emergency #preparedness starts with ...</td>\n      <td>1</td>\n      <td>1.0</td>\n      <td>0.623125</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"# most wronged predictions\nmost_wrong = val_df[val_df[\"target\"] != val_df[\"pred\"]].sort_values(\"pred_prob\", ascending=False)\nmost_wrong[:10]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:23.173969Z","iopub.execute_input":"2023-10-04T08:18:23.174621Z","iopub.status.idle":"2023-10-04T08:18:23.193713Z","shell.execute_reply.started":"2023-10-04T08:18:23.174580Z","shell.execute_reply":"2023-10-04T08:18:23.192395Z"},"trusted":true},"execution_count":208,"outputs":[{"execution_count":208,"output_type":"execute_result","data":{"text/plain":"                                                  text  target  pred  \\\n759  FedEx will no longer transport bioterror patho...       0   1.0   \n1    FedEx no longer to transport bioterror germs i...       0   1.0   \n49   @madonnamking RSPCA site multiple 7 story high...       0   1.0   \n549  Day 2. Liquidation of emergency at chemical ob...       0   1.0   \n209  Ashes 2015: AustraliaÛªs collapse at Trent Br...       0   1.0   \n344  Air Group is here to the rescue! We have 24/7 ...       0   1.0   \n360  Texas Seeks Comment on Rules for Changes to Wi...       0   1.0   \n31   ? High Skies - Burning Buildings ? http://t.co...       0   1.0   \n567  @RebeccaforReal accepts Wisconsin Emergency Re...       0   1.0   \n303  Trafford Centre film fans angry after Odeon ci...       0   1.0   \n\n     pred_prob  \n759   0.834415  \n1     0.770848  \n49    0.730688  \n549   0.711272  \n209   0.710154  \n344   0.698815  \n360   0.694015  \n31    0.688942  \n567   0.674546  \n303   0.671260  ","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>text</th>\n      <th>target</th>\n      <th>pred</th>\n      <th>pred_prob</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>759</th>\n      <td>FedEx will no longer transport bioterror patho...</td>\n      <td>0</td>\n      <td>1.0</td>\n      <td>0.834415</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>FedEx no longer to transport bioterror germs i...</td>\n      <td>0</td>\n      <td>1.0</td>\n      <td>0.770848</td>\n    </tr>\n    <tr>\n      <th>49</th>\n      <td>@madonnamking RSPCA site multiple 7 story high...</td>\n      <td>0</td>\n      <td>1.0</td>\n      <td>0.730688</td>\n    </tr>\n    <tr>\n      <th>549</th>\n      <td>Day 2. Liquidation of emergency at chemical ob...</td>\n      <td>0</td>\n      <td>1.0</td>\n      <td>0.711272</td>\n    </tr>\n    <tr>\n      <th>209</th>\n      <td>Ashes 2015: AustraliaÛªs collapse at Trent Br...</td>\n      <td>0</td>\n      <td>1.0</td>\n      <td>0.710154</td>\n    </tr>\n    <tr>\n      <th>344</th>\n      <td>Air Group is here to the rescue! We have 24/7 ...</td>\n      <td>0</td>\n      <td>1.0</td>\n      <td>0.698815</td>\n    </tr>\n    <tr>\n      <th>360</th>\n      <td>Texas Seeks Comment on Rules for Changes to Wi...</td>\n      <td>0</td>\n      <td>1.0</td>\n      <td>0.694015</td>\n    </tr>\n    <tr>\n      <th>31</th>\n      <td>? High Skies - Burning Buildings ? http://t.co...</td>\n      <td>0</td>\n      <td>1.0</td>\n      <td>0.688942</td>\n    </tr>\n    <tr>\n      <th>567</th>\n      <td>@RebeccaforReal accepts Wisconsin Emergency Re...</td>\n      <td>0</td>\n      <td>1.0</td>\n      <td>0.674546</td>\n    </tr>\n    <tr>\n      <th>303</th>\n      <td>Trafford Centre film fans angry after Odeon ci...</td>\n      <td>0</td>\n      <td>1.0</td>\n      <td>0.671260</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"# False positives Most Wrong\nfor row in most_wrong[:10].itertuples():\n    _, text, target, pred, prob = row\n    print(f\"Target: {target}, Pred: {int(pred)}, Prob: {prob}\")\n    print(f\"Text:\\n{text}\\n\")\n    print(\"----\\n\")","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:23.195675Z","iopub.execute_input":"2023-10-04T08:18:23.196179Z","iopub.status.idle":"2023-10-04T08:18:23.212013Z","shell.execute_reply.started":"2023-10-04T08:18:23.196136Z","shell.execute_reply":"2023-10-04T08:18:23.211013Z"},"trusted":true},"execution_count":209,"outputs":[{"name":"stdout","text":"Target: 0, Pred: 1, Prob: 0.8344152569770813\nText:\nFedEx will no longer transport bioterror pathogens in wake of anthrax lab mishaps http://t.co/lHpgxc4b8J\n\n----\n\nTarget: 0, Pred: 1, Prob: 0.7708482146263123\nText:\nFedEx no longer to transport bioterror germs in wake of anthrax lab mishaps http://t.co/qZQc8WWwcN via @usatoday\n\n----\n\nTarget: 0, Pred: 1, Prob: 0.7306876182556152\nText:\n@madonnamking RSPCA site multiple 7 story high rise buildings next to low density character residential in an area that floods\n\n----\n\nTarget: 0, Pred: 1, Prob: 0.7112718224525452\nText:\nDay 2. Liquidation of emergency at chemical object. #USAR2015 #USAR15 #RUOR #??????????? http://t.co/gGTmDqUdDo\n\n----\n\nTarget: 0, Pred: 1, Prob: 0.7101536989212036\nText:\nAshes 2015: AustraliaÛªs collapse at Trent Bridge among worst in history: England bundled out Australia for 60 ... http://t.co/t5TrhjUAU0\n\n----\n\nTarget: 0, Pred: 1, Prob: 0.6988154649734497\nText:\nAir Group is here to the rescue! We have 24/7 Emergency Service! Learn more about it here - http://t.co/9lyx7zMtHE http://t.co/5PbC96rTMJ\n\n----\n\nTarget: 0, Pred: 1, Prob: 0.6940151453018188\nText:\nTexas Seeks Comment on Rules for Changes to Windstorm Insurer http://t.co/BNNIdfZWbd\n\n----\n\nTarget: 0, Pred: 1, Prob: 0.6889419555664062\nText:\n? High Skies - Burning Buildings ? http://t.co/uVq41i3Kx2 #nowplaying\n\n----\n\nTarget: 0, Pred: 1, Prob: 0.6745464205741882\nText:\n@RebeccaforReal accepts Wisconsin Emergency Response Plan on behalf of @GovWalker #nbc15 http://t.co/Pis0aiVRbR\n\n----\n\nTarget: 0, Pred: 1, Prob: 0.6712601184844971\nText:\nTrafford Centre film fans angry after Odeon cinema evacuated following false fire alarm   http://t.co/6GLDwx71DA\n\n----\n\n","output_type":"stream"}]},{"cell_type":"code","source":"# false negatives\nfor row in most_wrong[-10:].itertuples():\n    _, text, target, pred, prob = row\n    print(f\"Target: {target}, Pred: {int(pred)}, Prob: {prob}\")\n    print(f\"Text:\\n{text}\\n\")\n    print(\"----\\n\")","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:23.213411Z","iopub.execute_input":"2023-10-04T08:18:23.214345Z","iopub.status.idle":"2023-10-04T08:18:23.228043Z","shell.execute_reply.started":"2023-10-04T08:18:23.214311Z","shell.execute_reply":"2023-10-04T08:18:23.226874Z"},"trusted":true},"execution_count":210,"outputs":[{"name":"stdout","text":"Target: 1, Pred: 0, Prob: 0.019319068640470505\nText:\n@GodOf_Mischief_ -of Loki's daggers she pulled it out and jammed it into Mina's thigh. When Mina screamed and grabbed at her leg sif-\n\n----\n\nTarget: 1, Pred: 0, Prob: 0.009843464940786362\nText:\n@DavidVonderhaar At least you were sincere ??\n\n----\n\nTarget: 1, Pred: 0, Prob: 0.00014119967818260193\nText:\n@willienelson We need help! Horses will die!Please RT &amp; sign petition!Take a stand &amp; be a voice for them! #gilbert23 https://t.co/e8dl1lNCVu\n\n----\n\nTarget: 1, Pred: 0, Prob: 0.0\nText:\n@SoonerMagic_ I mean I'm a fan but I don't need a girl sounding off like a damn siren\n\n----\n\nTarget: 1, Pred: 0, Prob: 0.0\nText:\nIf I fall is men GOD @Praiz8 is d bomb well av always known dat since 2008 bigger u I pray sir\n\n----\n\nTarget: 1, Pred: 0, Prob: 0.0\nText:\nReddit Will Now QuarantineÛ_ http://t.co/pkUAMXw6pm #onlinecommunities #reddit #amageddon #freespeech #Business http://t.co/PAWvNJ4sAP\n\n----\n\nTarget: 1, Pred: 0, Prob: 0.0\nText:\nIndeed!! I am fully aware of that battle! I support you in that fight!!  https://t.co/MctJnZX4H8\n\n----\n\nTarget: 1, Pred: 0, Prob: 0.0\nText:\ngoing to redo my nails and watch behind the scenes of desolation of smaug ayyy\n\n----\n\nTarget: 1, Pred: 0, Prob: 0.0\nText:\nRon &amp; Fez - Dave's High School Crush https://t.co/aN3W16c8F6 via @YouTube\n\n----\n\nTarget: 1, Pred: 0, Prob: 0.0\nText:\nI get to smoke my shit in peace\n\n----\n\n","output_type":"stream"}]},{"cell_type":"markdown","source":"## Making Prediction On Test Data","metadata":{}},{"cell_type":"code","source":"test_df.head()","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:23.229191Z","iopub.execute_input":"2023-10-04T08:18:23.229490Z","iopub.status.idle":"2023-10-04T08:18:23.248372Z","shell.execute_reply.started":"2023-10-04T08:18:23.229465Z","shell.execute_reply":"2023-10-04T08:18:23.247183Z"},"trusted":true},"execution_count":211,"outputs":[{"execution_count":211,"output_type":"execute_result","data":{"text/plain":"   id keyword location                                               text\n0   0     NaN      NaN                 Just happened a terrible car crash\n1   2     NaN      NaN  Heard about #earthquake is different cities, s...\n2   3     NaN      NaN  there is a forest fire at spot pond, geese are...\n3   9     NaN      NaN           Apocalypse lighting. #Spokane #wildfires\n4  11     NaN      NaN      Typhoon Soudelor kills 28 in China and Taiwan","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>id</th>\n      <th>keyword</th>\n      <th>location</th>\n      <th>text</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>0</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>Just happened a terrible car crash</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>2</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>Heard about #earthquake is different cities, s...</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>3</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>there is a forest fire at spot pond, geese are...</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>9</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>Apocalypse lighting. #Spokane #wildfires</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>11</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>Typhoon Soudelor kills 28 in China and Taiwan</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"# making prediction on test data\ntest_sentences = test_df[\"text\"].to_list()\ntest_samples = random.sample(test_sentences, 10)\n\nfor test_sample in test_samples:\n    pred_prob = tf.squeeze(model_6.predict([test_sample]))\n    pred = tf.round(pred_prob)\n    print(f\"Pred: {int(pred)}, Prob: {pred_prob}\")\n    print(f\"Text:\\n{test_sample}\\n\")\n    print(\"----\\n\")\n  ","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:23.250461Z","iopub.execute_input":"2023-10-04T08:18:23.250849Z","iopub.status.idle":"2023-10-04T08:18:24.140342Z","shell.execute_reply.started":"2023-10-04T08:18:23.250785Z","shell.execute_reply":"2023-10-04T08:18:24.139534Z"},"trusted":true},"execution_count":212,"outputs":[{"name":"stdout","text":"1/1 [==============================] - 0s 43ms/step\nPred: 0, Prob: 0.12780439853668213\nText:\nFirst responders tell me he is lucky.... had someone riding with him today. http://t.co/jc16CZn0NC\n\n----\n\n1/1 [==============================] - 0s 43ms/step\nPred: 0, Prob: 0.06662888079881668\nText:\nCreated save #666 on my current Fallout 3 playthrough. Hail Satan!\n\n----\n\n1/1 [==============================] - 0s 41ms/step\nPred: 1, Prob: 0.5793657898902893\nText:\nThis dude just pulled in front of our production with his truck on fire. Cops pulled up after we put it out http://t.co/e7DRPQUW4G\n\n----\n\n1/1 [==============================] - 0s 40ms/step\nPred: 1, Prob: 0.5264661908149719\nText:\nThe dust storm chase scene is pretty damn cool. #GhostProtocol\n\n----\n\n1/1 [==============================] - 0s 40ms/step\nPred: 1, Prob: 1.0170738697052002\nText:\n#?? #?? #??? #??? The Latest: More Homes Razed by Northern California Wildfire - New York Times  http://t.co/IPnHSIyorc-\n\n----\n\n1/1 [==============================] - 0s 43ms/step\nPred: 1, Prob: 1.1690219640731812\nText:\nGiant Sinkhole Swallows NYC Intersection http://t.co/ozIZdsDWP4 via @NewsBeatSocial\n\n----\n\n1/1 [==============================] - 0s 41ms/step\nPred: 1, Prob: 0.8495076894760132\nText:\nArsonist torches a house himself and his getaway vehicle police say - Surrey Leader http://t.co/q0J5lTVD6k @RichardGEarl\n\n----\n\n1/1 [==============================] - 0s 44ms/step\nPred: 0, Prob: 0.23579645156860352\nText:\n'imagine an entire aisle dedicated to making people look like serbian refugees.' - director of whole foods clothing section\n\n----\n\n1/1 [==============================] - 0s 43ms/step\nPred: 1, Prob: 0.756858766078949\nText:\n@TA7LPlays @Twitch_Bang @Maliceqt @Zelse007 @RhyareeGaming @dougiedabandit sorry what you say theres a tweet tsunami happening\n\n----\n\n1/1 [==============================] - 0s 42ms/step\nPred: 1, Prob: 0.9604756832122803\nText:\nShot 12 times. Found dead in cuffs after being involved in a car accident. Officers told ambulance not to treat him. https://t.co/MEUDJwaaNg\n\n----\n\n","output_type":"stream"}]},{"cell_type":"code","source":"test_df.head()","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:24.141619Z","iopub.execute_input":"2023-10-04T08:18:24.141972Z","iopub.status.idle":"2023-10-04T08:18:24.153038Z","shell.execute_reply.started":"2023-10-04T08:18:24.141943Z","shell.execute_reply":"2023-10-04T08:18:24.151824Z"},"trusted":true},"execution_count":213,"outputs":[{"execution_count":213,"output_type":"execute_result","data":{"text/plain":"   id keyword location                                               text\n0   0     NaN      NaN                 Just happened a terrible car crash\n1   2     NaN      NaN  Heard about #earthquake is different cities, s...\n2   3     NaN      NaN  there is a forest fire at spot pond, geese are...\n3   9     NaN      NaN           Apocalypse lighting. #Spokane #wildfires\n4  11     NaN      NaN      Typhoon Soudelor kills 28 in China and Taiwan","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>id</th>\n      <th>keyword</th>\n      <th>location</th>\n      <th>text</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>0</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>Just happened a terrible car crash</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>2</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>Heard about #earthquake is different cities, s...</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>3</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>there is a forest fire at spot pond, geese are...</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>9</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>Apocalypse lighting. #Spokane #wildfires</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>11</td>\n      <td>NaN</td>\n      <td>NaN</td>\n      <td>Typhoon Soudelor kills 28 in China and Taiwan</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"pred_prob = tf.squeeze(model_6.predict(test_df[\"text\"].to_numpy()))\npred = tf.round(pred_prob)\npred_prob[:10]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:24.154693Z","iopub.execute_input":"2023-10-04T08:18:24.155428Z","iopub.status.idle":"2023-10-04T08:18:25.061412Z","shell.execute_reply.started":"2023-10-04T08:18:24.155396Z","shell.execute_reply":"2023-10-04T08:18:25.060243Z"},"trusted":true},"execution_count":214,"outputs":[{"name":"stdout","text":"102/102 [==============================] - 1s 8ms/step\n","output_type":"stream"},{"execution_count":214,"output_type":"execute_result","data":{"text/plain":"<tf.Tensor: shape=(10,), dtype=float32, numpy=\narray([0.52173406, 0.84867346, 0.86173   , 0.96525025, 1.1235434 ,\n       0.64909047, 0.01546113, 0.        , 0.        , 0.        ],\n      dtype=float32)>"},"metadata":{}}]},{"cell_type":"code","source":"pred[:10]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:25.062544Z","iopub.execute_input":"2023-10-04T08:18:25.063151Z","iopub.status.idle":"2023-10-04T08:18:25.071279Z","shell.execute_reply.started":"2023-10-04T08:18:25.063119Z","shell.execute_reply":"2023-10-04T08:18:25.070301Z"},"trusted":true},"execution_count":215,"outputs":[{"execution_count":215,"output_type":"execute_result","data":{"text/plain":"<tf.Tensor: shape=(10,), dtype=float32, numpy=array([1., 1., 1., 1., 1., 1., 0., 0., 0., 0.], dtype=float32)>"},"metadata":{}}]},{"cell_type":"code","source":"test_data = test_df[:]","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:25.073006Z","iopub.execute_input":"2023-10-04T08:18:25.073335Z","iopub.status.idle":"2023-10-04T08:18:25.088725Z","shell.execute_reply.started":"2023-10-04T08:18:25.073308Z","shell.execute_reply":"2023-10-04T08:18:25.087410Z"},"trusted":true},"execution_count":216,"outputs":[]},{"cell_type":"code","source":"submission = test_df[['id']].reset_index(drop=True)\nsubmission['target'] = tf.cast(pred, tf.int64)\nsubmission","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:25.090459Z","iopub.execute_input":"2023-10-04T08:18:25.091100Z","iopub.status.idle":"2023-10-04T08:18:25.117250Z","shell.execute_reply.started":"2023-10-04T08:18:25.091060Z","shell.execute_reply":"2023-10-04T08:18:25.115740Z"},"trusted":true},"execution_count":217,"outputs":[{"execution_count":217,"output_type":"execute_result","data":{"text/plain":"         id  target\n0         0       1\n1         2       1\n2         3       1\n3         9       1\n4        11       1\n...     ...     ...\n3258  10861       1\n3259  10865       1\n3260  10868       1\n3261  10874       0\n3262  10875       1\n\n[3263 rows x 2 columns]","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>id</th>\n      <th>target</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>2</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>3</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>9</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>11</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>...</th>\n      <td>...</td>\n      <td>...</td>\n    </tr>\n    <tr>\n      <th>3258</th>\n      <td>10861</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>3259</th>\n      <td>10865</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>3260</th>\n      <td>10868</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>3261</th>\n      <td>10874</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3262</th>\n      <td>10875</td>\n      <td>1</td>\n    </tr>\n  </tbody>\n</table>\n<p>3263 rows × 2 columns</p>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"submission.to_csv('submission.csv', index=False)","metadata":{"execution":{"iopub.status.busy":"2023-10-04T08:18:25.119337Z","iopub.execute_input":"2023-10-04T08:18:25.120174Z","iopub.status.idle":"2023-10-04T08:18:25.132085Z","shell.execute_reply.started":"2023-10-04T08:18:25.120126Z","shell.execute_reply":"2023-10-04T08:18:25.130778Z"},"trusted":true},"execution_count":218,"outputs":[]}]}