automated_essay_scoring.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "automated_essay_scoring.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyONToysql9cloZQ7+dEQ3fI",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/Gail529/AES/blob/main/automated_essay_scoring.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "O8holrnjqoeb"
      },
      "source": [
        "# Importing  Necessary Modules"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BTILa78Y3xVM",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "d01961e7-2687-44a2-e886-9afe5e51c9f0"
      },
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "import nltk\n",
        "from nltk.corpus import stopwords\n",
        "from nltk.tokenize import sent_tokenize\n",
        " \n",
        "import re\n",
        "import string\n",
        "import warnings\n",
        "warnings.filterwarnings(\"ignore\")  \n",
        "from gensim.models import Word2Vec\n",
        " \n",
        " \n",
        "#essay preprocessing \n",
        "import nltk\n",
        "nltk.download('punkt')\n",
        "nltk.download('stopwords')\n",
        "nltk.download('wordnet')\n",
        "import string\n",
        "import re\n",
        "from nltk.tokenize import word_tokenize,sent_tokenize\n",
        "from nltk.stem import WordNetLemmatizer\n",
        "from nltk.corpus import stopwords\n",
        "from nltk.stem import PorterStemmer\n",
        " \n",
        "#training the neural network \n",
        "from tensorflow.keras.layers import Embedding\n",
        "from keras.models import Sequential, load_model, model_from_config\n",
        "from tensorflow.keras.preprocessing.sequence import pad_sequences\n",
        "from tensorflow.keras.models import Sequential\n",
        "from tensorflow.keras.layers import LSTM\n",
        "from tensorflow.keras.layers import Dropout\n",
        "from tensorflow.keras.layers import Dense\n",
        "from tensorflow.keras.layers import Lambda\n",
        "from tensorflow.keras.layers import Flatten\n",
        " \n",
        "from sklearn.linear_model import LinearRegression\n",
        "from sklearn.metrics import cohen_kappa_score\n",
        "import keras.backend as K"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[nltk_data] Downloading package punkt to /root/nltk_data...\n",
            "[nltk_data]   Unzipping tokenizers/punkt.zip.\n",
            "[nltk_data] Downloading package stopwords to /root/nltk_data...\n",
            "[nltk_data]   Unzipping corpora/stopwords.zip.\n",
            "[nltk_data] Downloading package wordnet to /root/nltk_data...\n",
            "[nltk_data]   Unzipping corpora/wordnet.zip.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WtpWZ4Vyqmhn"
      },
      "source": [
        "Preprocessing"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 107
        },
        "id": "1_-bDXNbIUpe",
        "outputId": "74cf5d5c-ff75-41bc-8c64-0489cc046b6e"
      },
      "source": [
        "#reading data set\n",
        "essay_data= pd.read_excel('/content/training_set_rel3.xlsx',usecols=['essay_id','essay','domain1_score'])\n",
        "essay_data.head(2)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "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>essay_id</th>\n",
              "      <th>essay</th>\n",
              "      <th>domain1_score</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1</td>\n",
              "      <td>Dear local newspaper, I think effects computer...</td>\n",
              "      <td>8.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>2</td>\n",
              "      <td>Dear @CAPS1 @CAPS2, I believe that using compu...</td>\n",
              "      <td>9.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   essay_id                                              essay  domain1_score\n",
              "0         1  Dear local newspaper, I think effects computer...            8.0\n",
              "1         2  Dear @CAPS1 @CAPS2, I believe that using compu...            9.0"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Xpkl7yetIdWD",
        "outputId": "64ac3019-6146-4062-d939-880deffa3c8b"
      },
      "source": [
        "x=essay_data['essay']\n",
        "y=essay_data['domain1_score']\n",
        "len(x) "
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "12978"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mB5IxYDird9g",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "cccbb384-4506-4277-a1ec-ca3a06d33bb1"
      },
      "source": [
        "#missing values\n",
        "essay_data['domain1_score'] = essay_data['domain1_score'].fillna(essay_data['domain1_score'].mean())\n",
        "print(essay_data.isnull().sum().sum())\n",
        "print(np.any(np.isnan(y).sum()))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0\n",
            "False\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gM5Xp0rbEaC_"
      },
      "source": [
        "#preprocessing the essays\n",
        "def sent_tokenize(text):\n",
        "    stripped_essay = text.strip()\n",
        "    #tokenization \n",
        "    tokenizer = nltk.data.load('tokenizers/punkt/english.pickle')\n",
        "    raw_sentences = tokenizer.tokenize(stripped_essay)\n",
        "    \n",
        "    tokenized_sentences = []\n",
        "    for raw_sentence in raw_sentences:\n",
        "        #removing punctuations and non alphanumeric characters\n",
        "        clean_sentence = re.sub(\"[^a-zA-Z0-9]\",\" \", raw_sentence)\n",
        "        #lowercasing\n",
        "        words = clean_sentence.lower().split()\n",
        "        #removing stopwords\n",
        "        stops = set(stopwords.words(\"english\"))\n",
        "        words = [w for w in words if not w in stops]\n",
        "        tokenized_sentences.append(words)\n",
        "\n",
        "    return tokenized_sentences\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "l7UIo0SnAWma"
      },
      "source": [
        "#sentences for the word2Vec model\n",
        "sentences=[]\n",
        "for essay in x:\n",
        "    sentences += sent_tokenize(essay)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-rKwV3LzB1Mt",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "cafacf77-51a8-4724-e7d2-712e65c85a4f"
      },
      "source": [
        "sentences[100]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['result', 'computers', 'would', 'improve', 'way', 'learn']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Pt1qTaNQ_Dva"
      },
      "source": [
        "# Word Embedding"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rvMn_Q2JLeTd",
        "outputId": "bf273c0d-9958-4364-bb97-da8b2277520d"
      },
      "source": [
        "text_dim = 300\n",
        "print(\"Training Word2Vec model...\")\n",
        "wordvec_model = Word2Vec(sentences, size=text_dim, window=5, min_count=3, workers=4, sg=1)\n",
        "print(\"Word2Vec model created.\")\n",
        "print(\"%d unique words represented by %d dimensional vectors\" % (len(wordvec_model.wv.vocab), text_dim))\n",
        "wordvec_model.save('wordvec_model')\n",
        "print(\"Word2Vec model saved.\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Training Word2Vec model...\n",
            "Word2Vec model created.\n",
            "13134 unique words represented by 300 dimensional vectors\n",
            "Word2Vec model saved.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "es33b9LNUUla"
      },
      "source": [
        "#essays\n",
        "cleaned_essays=[]\n",
        "for i in range(0,len(x)):\n",
        "    essay=x[i]\n",
        "    cleaned_essays.append(sent_tokenize(essay))\n",
        "    "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7gaRkCN1TBy-"
      },
      "source": [
        "def clean_essays(essay):\n",
        "    clean_essay= re.sub(\"[^a-zA-Z]\", \" \",essay)\n",
        "    words = essay.lower().split()\n",
        "    stops = set(stopwords.words(\"english\"))\n",
        "    words = [w for w in words if not w in stops]\n",
        "    return (words)\n",
        " \n",
        "cleaned_essays = []\n",
        "for essay in x:\n",
        "        cleaned_essays.append(clean_essays(essay))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nBIOCDifPoMk",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "d0b5595b-4c99-400b-feba-21923e488b07"
      },
      "source": [
        "#creating the embedding matrix\n",
        "def create_average_vec(essay):\n",
        "    average = np.zeros((text_dim,), dtype='float32')\n",
        "    num_words = 0.\n",
        "    for word in essay:\n",
        "        if word in wordvec_model.wv.vocab:\n",
        "            average = np.add(average, wordvec_model.wv[word])\n",
        "            num_words += 1.\n",
        "    if num_words != 0.:\n",
        "        average = np.divide(average, num_words)\n",
        "    return average\n",
        "\n",
        "cleaned_vec = np.zeros((len(x),text_dim), dtype=\"float32\")  \n",
        "for i in range(len(cleaned_essays)):\n",
        "    cleaned_vec[i] = create_average_vec(cleaned_essays[i])\n",
        "\n",
        "print(\"Word vectors for all essays in the training data set are of shape:\", cleaned_vec.shape)\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Word vectors for all essays in the training data set are of shape: (12978, 300)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "23_vYXi1ibA9",
        "outputId": "903a2456-390e-47fc-bdaf-e3494cc0ff9f"
      },
      "source": [
        "#splitting the training and validation set\n",
        "data_split=0.8\n",
        "train_samples=int(len(cleaned_vec)*data_split)\n",
        "\n",
        "x_train = np.array(cleaned_vec[:train_samples])\n",
        "x_test = np.array(cleaned_vec[train_samples:])\n",
        "y_train = np.array(y[:train_samples])\n",
        "y_test = np.array(y[train_samples:])\n",
        "\n",
        "print(\"x_train\" ,len(x_train))\n",
        "print(\"x_test\" ,len(x_test))\n",
        "print(\"y_train\" ,len(y_train))\n",
        "print(\"y_test\" ,len(y_test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "x_train 10382\n",
            "x_test 2596\n",
            "y_train 10382\n",
            "y_test 2596\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1_16xPRvcbD5"
      },
      "source": [
        "#playing around with the hyperparameters\n",
        "def get_model():\n",
        "    model = Sequential()\n",
        "    model.add(LSTM(300, dropout=0.4, recurrent_dropout=0.4, input_shape=[1, 300], return_sequences=True))\n",
        "    model.add(LSTM(64, recurrent_dropout=0.4))\n",
        "    model.add(Dropout(0.5))\n",
        "    model.add(Dense(1, activation='tanh'))\n",
        "    model.compile(loss='mean_squared_error', optimizer='adam', metrics=['accuracy'])\n",
        "    model.summary()\n",
        "    return model\n",
        "\n",
        "def get_model2():\n",
        "    model = Sequential()\n",
        "    model.add(LSTM(300, dropout=0.4, recurrent_dropout=0.4, input_shape=[1, 300], return_sequences=True))\n",
        "    model.add(LSTM(64, recurrent_dropout=0.4))\n",
        "    model.add(Dropout(0.5))\n",
        "    model.add(Dense(1, activation='relu'))\n",
        "    model.compile(loss='mean_squared_error', optimizer='rmsprop', metrics=['mae'])\n",
        "    model.summary()\n",
        "    return model\n",
        "\n",
        "def get_model3():\n",
        "    model = Sequential()\n",
        "    model.add(LSTM(300, dropout=0.4, recurrent_dropout=0.4, input_shape=[1, 300], return_sequences=True))\n",
        "    model.add(LSTM(64, recurrent_dropout=0.4))\n",
        "    model.add(Dropout(0.5))\n",
        "    model.add(Dense(1, activation='tanh'))\n",
        "    model.compile(loss='mean_squared_error', optimizer='adam', metrics=['mae'])\n",
        "    model.summary()\n",
        "    return model\n",
        "\n",
        "def get_model4():\n",
        "    model = Sequential()\n",
        "    model.add(LSTM(300, dropout=0.4, recurrent_dropout=0.4, input_shape=[1, 300], return_sequences=True))\n",
        "    model.add(LSTM(64, recurrent_dropout=0.4))\n",
        "    model.add(Dropout(0.5))\n",
        "    model.add(Dense(1, activation='tanh'))\n",
        "    model.compile(loss='mean_squared_error', optimizer='rmsprop', metrics=['mae'])\n",
        "    model.summary()\n",
        "    return model\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JHIzH4lncdl3",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "0b57cb23-c0b1-4f93-aac9-b8c24d3b17d0"
      },
      "source": [
        "x_train = np.reshape(x_train, (x_train.shape[0], 1, x_train.shape[1]))\n",
        "x_test = np.reshape(x_test, (x_test.shape[0], 1, x_test.shape[1]))\n",
        "    \n",
        " \n",
        "lstm_model = get_model3()\n",
        "history=lstm_model.fit(x_train, y_train, batch_size=100, epochs=20,validation_data=(x_test, y_test))\n",
        " \n",
        "y_pred = lstm_model.predict(x_test)\n",
        "lstm_model.save('final_lstm.h5')\n",
        " \n",
        "# Round y_pred to the nearest integer.\n",
        "y_pred = np.around(y_pred)\n",
        " \n",
        "# Evaluate the model on the evaluation metric. \"Quadratic mean averaged Kappa\"\n",
        "result = cohen_kappa_score(y_test,y_pred,weights='quadratic')\n",
        "print(\"Kappa Score: {}\".format(result))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_3\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "lstm_6 (LSTM)                (None, 1, 300)            721200    \n",
            "_________________________________________________________________\n",
            "lstm_7 (LSTM)                (None, 64)                93440     \n",
            "_________________________________________________________________\n",
            "dropout_3 (Dropout)          (None, 64)                0         \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 1)                 65        \n",
            "=================================================================\n",
            "Total params: 814,705\n",
            "Trainable params: 814,705\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Epoch 1/20\n",
            "104/104 [==============================] - 9s 38ms/step - loss: 14.3417 - mae: 2.7420 - val_loss: 519.4193 - val_mae: 19.3189\n",
            "Epoch 2/20\n",
            "104/104 [==============================] - 3s 33ms/step - loss: 12.2895 - mae: 2.4458 - val_loss: 519.4188 - val_mae: 19.3189\n",
            "Epoch 3/20\n",
            "104/104 [==============================] - 3s 33ms/step - loss: 12.8389 - mae: 2.5161 - val_loss: 519.4184 - val_mae: 19.3189\n",
            "Epoch 4/20\n",
            "104/104 [==============================] - 3s 33ms/step - loss: 12.7643 - mae: 2.5045 - val_loss: 519.4180 - val_mae: 19.3189\n",
            "Epoch 5/20\n",
            "104/104 [==============================] - 4s 34ms/step - loss: 12.7900 - mae: 2.4938 - val_loss: 519.4178 - val_mae: 19.3189\n",
            "Epoch 6/20\n",
            "104/104 [==============================] - 3s 33ms/step - loss: 12.4978 - mae: 2.4684 - val_loss: 519.4177 - val_mae: 19.3189\n",
            "Epoch 7/20\n",
            "104/104 [==============================] - 3s 33ms/step - loss: 12.6167 - mae: 2.4902 - val_loss: 519.4175 - val_mae: 19.3189\n",
            "Epoch 8/20\n",
            "104/104 [==============================] - 3s 33ms/step - loss: 12.9679 - mae: 2.5170 - val_loss: 519.4174 - val_mae: 19.3189\n",
            "Epoch 9/20\n",
            "104/104 [==============================] - 3s 32ms/step - loss: 12.9260 - mae: 2.5119 - val_loss: 519.4174 - val_mae: 19.3189\n",
            "Epoch 10/20\n",
            "104/104 [==============================] - 3s 33ms/step - loss: 13.1621 - mae: 2.5505 - val_loss: 519.4173 - val_mae: 19.3189\n",
            "Epoch 11/20\n",
            "104/104 [==============================] - 3s 32ms/step - loss: 12.6614 - mae: 2.4970 - val_loss: 519.4172 - val_mae: 19.3189\n",
            "Epoch 12/20\n",
            "104/104 [==============================] - 3s 33ms/step - loss: 12.9594 - mae: 2.5325 - val_loss: 519.4171 - val_mae: 19.3189\n",
            "Epoch 13/20\n",
            "104/104 [==============================] - 3s 33ms/step - loss: 12.3517 - mae: 2.4661 - val_loss: 519.4171 - val_mae: 19.3189\n",
            "Epoch 14/20\n",
            "104/104 [==============================] - 3s 33ms/step - loss: 12.6481 - mae: 2.4914 - val_loss: 519.4171 - val_mae: 19.3189\n",
            "Epoch 15/20\n",
            "104/104 [==============================] - 3s 33ms/step - loss: 12.5447 - mae: 2.4820 - val_loss: 519.4171 - val_mae: 19.3189\n",
            "Epoch 16/20\n",
            "104/104 [==============================] - 3s 32ms/step - loss: 12.5296 - mae: 2.4713 - val_loss: 519.4171 - val_mae: 19.3189\n",
            "Epoch 17/20\n",
            "104/104 [==============================] - 3s 33ms/step - loss: 12.7982 - mae: 2.5144 - val_loss: 519.4171 - val_mae: 19.3189\n",
            "Epoch 18/20\n",
            "104/104 [==============================] - 3s 33ms/step - loss: 12.6880 - mae: 2.4826 - val_loss: 519.4171 - val_mae: 19.3189\n",
            "Epoch 19/20\n",
            "104/104 [==============================] - 3s 33ms/step - loss: 12.9744 - mae: 2.5225 - val_loss: 519.4171 - val_mae: 19.3189\n",
            "Epoch 20/20\n",
            "104/104 [==============================] - 3s 33ms/step - loss: 12.8803 - mae: 2.5181 - val_loss: 519.4170 - val_mae: 19.3189\n",
            "Kappa Score: 2.5698828580722477e-05\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 294
        },
        "id": "2ZafpXpQ40HN",
        "outputId": "cdb1f149-a124-4d20-be3d-d222bf1a4952"
      },
      "source": [
        "plt.plot(history.history['loss'], label='MAE (training data)')\n",
        "plt.plot(history.history['val_loss'], label='MAE (validation data)')\n",
        "plt.title('MAE')\n",
        "plt.ylabel('MAE value')\n",
        "plt.xlabel('No. epoch')\n",
        "plt.legend(loc=\"upper left\")\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IK5dUTN62lFp",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 350
        },
        "outputId": "db657e3e-16c7-4812-b642-a0aea6c45c4f"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "\n",
        "loss= history.history['loss']\n",
        "val_loss=history.history['val_loss']\n",
        "\n",
        "acc=history.history['accuracy']\n",
        "val_acc = history.history['val_accuracy']\n",
        "\n",
        "epochs = range(1,len(loss)+1)\n",
        "fig, (ax1,ax2) = plt.subplots(1,2,figsize=(15,5))\n",
        "\n",
        "ax1.plot(epochs,loss,label='Training loss')\n",
        "ax1.plot(epochs,val_loss)\n",
        "ax1.set_title('Training and Validation loss')\n",
        "ax1.set_xlabel('Epochs')\n",
        "ax1.set_ylabel('Loss')\n",
        "ax1.legend()\n",
        "\n",
        "\n",
        "ax2.plot(epochs,loss,label='Training Accuracy')\n",
        "ax2.plot(epochs,val_acc,label='Validation Accuracy')\n",
        "ax2.set_title('Training and Validation accuracy')\n",
        "ax2.set_xlabel('Epochs')\n",
        "ax2.set_ylabel('Accuracy')\n",
        "ax2.legend()\n",
        "\n",
        "\n",
        "plt.show()\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x360 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "s-brmjdSsW07",
        "outputId": "4aea860a-d90b-4ab4-9608-460f8d553ead"
      },
      "source": [
        "x_train = np.reshape(x_train, (x_train.shape[0], 1, x_train.shape[1]))\n",
        "x_test = np.reshape(x_test, (x_test.shape[0], 1, x_test.shape[1]))\n",
        "    \n",
        " \n",
        "lstm_model = get_model2()\n",
        "lstm_model.fit(x_train, y_train, batch_size=100, epochs=20)\n",
        "y_pred = lstm_model.predict(x_test)\n",
        "lstm_model.save('final_lstm2.h5')\n",
        " \n",
        "# Round y_pred to the nearest integer.\n",
        "y_pred = np.around(y_pred)\n",
        " \n",
        "# Evaluate the model on the evaluation metric. \"Quadratic mean averaged Kappa\"\n",
        "result = cohen_kappa_score(y_test,y_pred,weights='quadratic')\n",
        "print(\"Kappa Score: {}\".format(result))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "lstm_2 (LSTM)                (None, 1, 300)            721200    \n",
            "_________________________________________________________________\n",
            "lstm_3 (LSTM)                (None, 64)                93440     \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 64)                0         \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 1)                 65        \n",
            "=================================================================\n",
            "Total params: 814,705\n",
            "Trainable params: 814,705\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Epoch 1/20\n",
            "104/104 [==============================] - 8s 31ms/step - loss: 6.9695 - mae: 1.8185\n",
            "Epoch 2/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 2.2457 - mae: 1.1317\n",
            "Epoch 3/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 2.0415 - mae: 1.0507\n",
            "Epoch 4/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 1.8351 - mae: 0.9984\n",
            "Epoch 5/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 1.6934 - mae: 0.9725\n",
            "Epoch 6/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 1.6742 - mae: 0.9556\n",
            "Epoch 7/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 1.5857 - mae: 0.9226\n",
            "Epoch 8/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 1.4793 - mae: 0.9087\n",
            "Epoch 9/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 1.3904 - mae: 0.8824\n",
            "Epoch 10/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 1.3739 - mae: 0.8700\n",
            "Epoch 11/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 1.3032 - mae: 0.8606\n",
            "Epoch 12/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 1.3015 - mae: 0.8589\n",
            "Epoch 13/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 1.2486 - mae: 0.8333\n",
            "Epoch 14/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 1.2454 - mae: 0.8420\n",
            "Epoch 15/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 1.1784 - mae: 0.8183\n",
            "Epoch 16/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 1.1857 - mae: 0.8194\n",
            "Epoch 17/20\n",
            "104/104 [==============================] - 3s 29ms/step - loss: 1.1523 - mae: 0.8101\n",
            "Epoch 18/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 1.1840 - mae: 0.8131\n",
            "Epoch 19/20\n",
            "104/104 [==============================] - 3s 31ms/step - loss: 1.1208 - mae: 0.7926\n",
            "Epoch 20/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 1.1623 - mae: 0.7989\n",
            "Kappa Score: 0.014519219835083863\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "B9i7GUjicHgv",
        "outputId": "8610b74f-2406-45a4-96e2-9f9e96a67680"
      },
      "source": [
        "x_train = np.reshape(x_train, (x_train.shape[0], 1, x_train.shape[1]))\n",
        "x_test = np.reshape(x_test, (x_test.shape[0], 1, x_test.shape[1]))\n",
        "    \n",
        " \n",
        "lstm_model = get_model4()\n",
        "lstm_model.fit(x_train, y_train, batch_size=100, epochs=20)\n",
        "y_pred = lstm_model.predict(x_test)\n",
        "lstm_model.save('final_lstm4.h5')\n",
        " \n",
        "# Round y_pred to the nearest integer.\n",
        "y_pred = np.around(y_pred)\n",
        " \n",
        "# Evaluate the model on the evaluation metric. \"Quadratic mean averaged Kappa\"\n",
        "result = cohen_kappa_score(y_test,y_pred,weights='quadratic')\n",
        "print(\"Kappa Score: {}\".format(result))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_4\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "lstm_8 (LSTM)                (None, 1, 300)            721200    \n",
            "_________________________________________________________________\n",
            "lstm_9 (LSTM)                (None, 64)                93440     \n",
            "_________________________________________________________________\n",
            "dropout_4 (Dropout)          (None, 64)                0         \n",
            "_________________________________________________________________\n",
            "dense_4 (Dense)              (None, 1)                 65        \n",
            "=================================================================\n",
            "Total params: 814,705\n",
            "Trainable params: 814,705\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Epoch 1/20\n",
            "104/104 [==============================] - 7s 29ms/step - loss: 14.0811 - mae: 2.6886\n",
            "Epoch 2/20\n",
            "104/104 [==============================] - 3s 29ms/step - loss: 12.9203 - mae: 2.5200\n",
            "Epoch 3/20\n",
            "104/104 [==============================] - 3s 29ms/step - loss: 12.7562 - mae: 2.5074\n",
            "Epoch 4/20\n",
            "104/104 [==============================] - 3s 29ms/step - loss: 13.0556 - mae: 2.5309\n",
            "Epoch 5/20\n",
            "104/104 [==============================] - 3s 29ms/step - loss: 12.5212 - mae: 2.4859\n",
            "Epoch 6/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 12.6284 - mae: 2.4928\n",
            "Epoch 7/20\n",
            "104/104 [==============================] - 3s 29ms/step - loss: 12.5253 - mae: 2.4738\n",
            "Epoch 8/20\n",
            "104/104 [==============================] - 3s 29ms/step - loss: 12.7611 - mae: 2.4924\n",
            "Epoch 9/20\n",
            "104/104 [==============================] - 3s 29ms/step - loss: 12.4468 - mae: 2.4729\n",
            "Epoch 10/20\n",
            "104/104 [==============================] - 3s 29ms/step - loss: 12.5483 - mae: 2.4883\n",
            "Epoch 11/20\n",
            "104/104 [==============================] - 3s 29ms/step - loss: 13.0675 - mae: 2.5334\n",
            "Epoch 12/20\n",
            "104/104 [==============================] - 3s 29ms/step - loss: 12.8560 - mae: 2.5019\n",
            "Epoch 13/20\n",
            "104/104 [==============================] - 3s 29ms/step - loss: 12.8090 - mae: 2.5061\n",
            "Epoch 14/20\n",
            "104/104 [==============================] - 3s 29ms/step - loss: 12.5188 - mae: 2.4841\n",
            "Epoch 15/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 13.0227 - mae: 2.5296\n",
            "Epoch 16/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 12.7223 - mae: 2.5038\n",
            "Epoch 17/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 12.8763 - mae: 2.5255\n",
            "Epoch 18/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 12.7137 - mae: 2.4969\n",
            "Epoch 19/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 12.5405 - mae: 2.4759\n",
            "Epoch 20/20\n",
            "104/104 [==============================] - 3s 30ms/step - loss: 12.8727 - mae: 2.5094\n",
            "Kappa Score: 0.0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HDvCVua06m9N"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}