2024 工业大数据/制造关键装置故障诊断baseline_lgb.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-01-13T11:33:31.962824Z",
     "iopub.status.busy": "2024-01-13T11:33:31.962343Z",
     "iopub.status.idle": "2024-01-13T11:33:31.975250Z",
     "shell.execute_reply": "2024-01-13T11:33:31.974175Z",
     "shell.execute_reply.started": "2024-01-13T11:33:31.962788Z"
    }
   },
   "outputs": [],
   "source": [
    "import os\n",
    "import warnings\n",
    "import scipy.stats\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import lightgbm as lgb\n",
    "from tqdm.auto import tqdm\n",
    "from scipy.fftpack import fft\n",
    "from scipy.stats import skew, kurtosis\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.model_selection import StratifiedKFold,KFold,train_test_split\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-01-13T11:34:14.954187Z",
     "iopub.status.busy": "2024-01-13T11:34:14.953746Z",
     "iopub.status.idle": "2024-01-13T11:34:14.982306Z",
     "shell.execute_reply": "2024-01-13T11:34:14.980812Z",
     "shell.execute_reply.started": "2024-01-13T11:34:14.954152Z"
    }
   },
   "outputs": [],
   "source": [
    "def model_train_multiclassifier(df_train, df_test, feats, seed, label_name, n_fold=10):\n",
    "    skf = StratifiedKFold(n_splits=n_fold, shuffle=True, random_state=seed)\n",
    "    train_label = df_train[label_name]\n",
    "    label_num = int(train_label.max() + 1)\n",
    "    i = 0\n",
    "    importance = 0\n",
    "    pred_y = 0\n",
    "    oof = np.zeros([len(df_train), label_num])\n",
    "    pred_y = np.zeros([len(df_test), label_num])\n",
    "    params = {\n",
    "        'learning_rate': 0.05,\n",
    "        'boosting_type': 'gbdt',\n",
    "        'objective': 'multiclass',\n",
    "        'num_class': label_num,\n",
    "        'metric': 'multi_error',\n",
    "        'num_leaves': 8,\n",
    "        'verbose': -1,\n",
    "        'seed': 42,\n",
    "        'n_jobs': -1,\n",
    "        'feature_fraction': 0.5,\n",
    "        'bagging_fraction': 0.5,\n",
    "    }\n",
    "\n",
    "    for fold, (train_idx, val_idx) in enumerate(skf.split(df_train, train_label)):\n",
    "        print('---------------------------', fold)\n",
    "        train = lgb.Dataset(df_train.loc[train_idx, feats],\n",
    "                            df_train.loc[train_idx, label_name])\n",
    "        val = lgb.Dataset(df_train.loc[val_idx, feats],df_train.loc[val_idx, label_name])\n",
    "        model = lgb.train(params, train, valid_sets=val, num_boost_round=5000,\n",
    "                          callbacks=[lgb.early_stopping(100), lgb.log_evaluation(500)])\n",
    "        oof[val_idx] = model.predict(df_train.loc[val_idx, feats])\n",
    "        importance += model.feature_importance(importance_type='gain') / n_fold\n",
    "        pred_y += model.predict(df_test[feats].to_numpy()) / n_fold\n",
    "    feats_importance = pd.DataFrame()\n",
    "    feats_importance['name'] = feats\n",
    "    feats_importance['importance'] = importance\n",
    "    return pred_y, oof, feats_importance.sort_values(\"importance\", ascending=False)\n",
    "                  \n",
    "def extract_segmented_features(data, segment_size=256, start_point=128):\n",
    "    num_segments = (data.shape[1] - start_point) // segment_size\n",
    "                     \n",
    "    segmented_features_list = []\n",
    "    for i in range(num_segments):\n",
    "        start_idx = start_point + i * segment_size\n",
    "        end_idx = start_idx + segment_size\n",
    "        \n",
    "        segment_data = data[:, start_idx:end_idx]\n",
    "\n",
    "        mean = np.mean(segment_data, axis=1)\n",
    "        std = np.std(segment_data, axis=1)\n",
    "        median = np.median(segment_data, axis=1)\n",
    "        skewness = skew(segment_data, axis=1)\n",
    "        kurt = kurtosis(segment_data, axis=1)\n",
    "        \n",
    "        # 频域特征\n",
    "        fft_result = fft(segment_data, axis=1)\n",
    "        fft_magnitude = np.abs(fft_result)\n",
    "        max_freq_index = np.argmax(fft_magnitude, axis=1)\n",
    "        max_freq = np.fft.fftfreq(segment_data.shape[1])[max_freq_index]\n",
    "        echo_intervals = np.argmax(fft_result[:, 1:], axis=1) + 1  \n",
    "        features_dict = {\n",
    "            'mean': mean,\n",
    "            'std': std,\n",
    "            'median': median,\n",
    "            'skewness': skewness,\n",
    "            'kurtosis': kurt,\n",
    "            'max_freq': max_freq,\n",
    "        }\n",
    "\n",
    "        features_df = pd.DataFrame(features_dict)\n",
    "        segmented_features_list.append(features_df)\n",
    "    combined_features_df = pd.concat(segmented_features_list, axis=1)\n",
    "    feature_names = [f\"{start_point}_{segment_size}_{stat}_{i+1}\" for stat in features_dict.keys() for i in range(num_segments)]\n",
    "    combined_features_df.columns = feature_names\n",
    "    return combined_features_df,feature_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-01-13T11:29:10.042342Z",
     "iopub.status.busy": "2024-01-13T11:29:10.041776Z",
     "iopub.status.idle": "2024-01-13T11:29:19.261379Z",
     "shell.execute_reply": "2024-01-13T11:29:19.259999Z",
     "shell.execute_reply.started": "2024-01-13T11:29:10.042305Z"
    }
   },
   "outputs": [],
   "source": [
    "test_path=\"data/test1/\"\n",
    "dfs = []\n",
    "for i in range(0,2000):\n",
    "    file_path=test_path+str(i)+\".txt\"\n",
    "    df = pd.read_csv(file_path, sep='\\t', header=None)\n",
    "    df=df.T\n",
    "    dfs.append(df)\n",
    "\n",
    "df_test = pd.concat(dfs).reset_index(drop=True)\n",
    " "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-01-13T11:29:19.264511Z",
     "iopub.status.busy": "2024-01-13T11:29:19.264070Z",
     "iopub.status.idle": "2024-01-13T11:29:37.721921Z",
     "shell.execute_reply": "2024-01-13T11:29:37.719975Z",
     "shell.execute_reply.started": "2024-01-13T11:29:19.264474Z"
    }
   },
   "outputs": [],
   "source": [
    "df_all = []\n",
    "for i in range(4):\n",
    "    train_path=\"data/train/%d\"%i\n",
    "    dfs = []\n",
    "    for file_name in os.listdir(train_path):\n",
    "        if file_name.endswith('.txt'):\n",
    "            file_path = os.path.join(train_path, file_name)\n",
    "            df = pd.read_csv(file_path, sep='\\t', header=None)\n",
    "            dfs.append(df.T)\n",
    "\n",
    "    df_train = pd.concat(dfs)         \n",
    "    df_train[\"label\"]=i\n",
    "    df_all.append(df_train)\n",
    "\n",
    "df_train=pd.concat(df_all).reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-01-13T11:34:19.325581Z",
     "iopub.status.busy": "2024-01-13T11:34:19.325119Z",
     "iopub.status.idle": "2024-01-13T11:34:26.101485Z",
     "shell.execute_reply": "2024-01-13T11:34:26.099480Z",
     "shell.execute_reply.started": "2024-01-13T11:34:19.325552Z"
    }
   },
   "outputs": [],
   "source": [
    "feats = [i for i in df_test.columns]\n",
    "label = df_train[\"label\"]\n",
    "df_test_,feats_=extract_segmented_features(df_test[feats].to_numpy(),256,0)\n",
    "df_train_,feats_=extract_segmented_features(df_train[feats].to_numpy(),256,0)\n",
    "\n",
    "df_train_[\"label\"]=label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--------------------------- 0\n",
      "Training until validation scores don't improve for 100 rounds\n",
      "Early stopping, best iteration is:\n",
      "[223]\tvalid_0's multi_error: 0.03375\n",
      "--------------------------- 1\n",
      "Training until validation scores don't improve for 100 rounds\n",
      "Early stopping, best iteration is:\n",
      "[190]\tvalid_0's multi_error: 0.04625\n",
      "--------------------------- 2\n",
      "Training until validation scores don't improve for 100 rounds\n",
      "Early stopping, best iteration is:\n",
      "[198]\tvalid_0's multi_error: 0.06\n",
      "--------------------------- 3\n",
      "Training until validation scores don't improve for 100 rounds\n",
      "Early stopping, best iteration is:\n",
      "[315]\tvalid_0's multi_error: 0.04\n",
      "--------------------------- 4\n",
      "Training until validation scores don't improve for 100 rounds\n",
      "Early stopping, best iteration is:\n",
      "[159]\tvalid_0's multi_error: 0.04875\n"
     ]
    }
   ],
   "source": [
    "pred_y, oof, feats_importance=model_train_multiclassifier(df_train_,df_test_,feats_,22222,\"label\",5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-01-13T11:35:39.070757Z",
     "iopub.status.busy": "2024-01-13T11:35:39.069948Z",
     "iopub.status.idle": "2024-01-13T11:35:39.088717Z",
     "shell.execute_reply": "2024-01-13T11:35:39.087048Z",
     "shell.execute_reply.started": "2024-01-13T11:35:39.070698Z"
    }
   },
   "outputs": [
    {
     "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>name</th>\n",
       "      <th>importance</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>0_256_median_16</td>\n",
       "      <td>6793.619238</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>53</th>\n",
       "      <td>0_256_skewness_6</td>\n",
       "      <td>5240.415136</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0_256_mean_6</td>\n",
       "      <td>5189.007815</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>71</th>\n",
       "      <td>0_256_kurtosis_8</td>\n",
       "      <td>4547.275810</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>59</th>\n",
       "      <td>0_256_skewness_12</td>\n",
       "      <td>4245.360804</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>0_256_std_12</td>\n",
       "      <td>18.944722</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>0_256_median_14</td>\n",
       "      <td>15.995582</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0_256_mean_4</td>\n",
       "      <td>15.313492</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>0_256_median_8</td>\n",
       "      <td>10.939103</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>69</th>\n",
       "      <td>0_256_kurtosis_6</td>\n",
       "      <td>7.665566</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>96 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                 name   importance\n",
       "47    0_256_median_16  6793.619238\n",
       "53   0_256_skewness_6  5240.415136\n",
       "5        0_256_mean_6  5189.007815\n",
       "71   0_256_kurtosis_8  4547.275810\n",
       "59  0_256_skewness_12  4245.360804\n",
       "..                ...          ...\n",
       "27       0_256_std_12    18.944722\n",
       "45    0_256_median_14    15.995582\n",
       "3        0_256_mean_4    15.313492\n",
       "39     0_256_median_8    10.939103\n",
       "69   0_256_kurtosis_6     7.665566\n",
       "\n",
       "[96 rows x 2 columns]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feats_importance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-01-13T11:35:39.092108Z",
     "iopub.status.busy": "2024-01-13T11:35:39.090722Z",
     "iopub.status.idle": "2024-01-13T11:35:39.110661Z",
     "shell.execute_reply": "2024-01-13T11:35:39.109536Z",
     "shell.execute_reply.started": "2024-01-13T11:35:39.092046Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.95425\n"
     ]
    }
   ],
   "source": [
    "print(accuracy_score(label,np.argmax(oof,axis=1)))\n",
    "df_test[\"label\"]=np.argmax(pred_y,axis=1)\n",
    "df_test[[\"label\"]].to_csv(\"lgb_{:.4f}.csv\".format(accuracy_score(label,np.argmax(oof,axis=1))),header=None,index=False) "
   ]
  }
 ],
 "metadata": {
  "kaggle": {
   "accelerator": "none",
   "dataSources": [
    {
     "datasetId": 4285451,
     "sourceId": 7375181,
     "sourceType": "datasetVersion"
    }
   ],
   "dockerImageVersionId": 30626,
   "isGpuEnabled": false,
   "isInternetEnabled": true,
   "language": "python",
   "sourceType": "notebook"
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.13"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": false
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}