run.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from config import *\n",
    "from dataset import *\n",
    "from model import *\n",
    "from train import *\n",
    "from anomaly_score import *\n",
    "from evaluate import *\n",
    "\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "from torch.utils.data import DataLoader"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# DATA"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 데이터 불러 오기\n",
    "\n",
    "data_path = \"data/ptbdb_normal.csv\"\n",
    "data = pd.read_csv(data_path, names=[f\"sensor{i}\" for i in range(188)]).iloc[:, :16]\n",
    "\n",
    "split_size01 = int(data.shape[0] * 0.6)  # data split size\n",
    "split_size02 = int(data.shape[0] * 0.9)  # data split size\n",
    "\n",
    "# train:valid - 7:3\n",
    "train = data[:split_size01]\n",
    "valid = data[split_size01:split_size02]\n",
    "threshold_normal = data[split_size02:]\n",
    "\n",
    "# abnormal data\n",
    "data_path = \"data/ptbdb_abnormal.csv\"\n",
    "data = pd.read_csv(data_path, names=[f\"sensor{i}\" for i in range(188)]).iloc[:, :16]\n",
    "\n",
    "split_size = int(data.shape[0] * 0.3)  # data split size\n",
    "\n",
    "# train:valid - 7:3\n",
    "threshold_abnormal = data[:split_size]\n",
    "test = data[split_size:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "config = Config().config  # load config"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "정규화 완료\n",
      "정규화 완료\n",
      "정규화 완료\n",
      "정규화 완료\n",
      "정규화 완료\n"
     ]
    }
   ],
   "source": [
    "train_dataset = AD_Dataset(data=train.values, config=config)\n",
    "valid_dataset = AD_Dataset(data=valid.values, config=config)\n",
    "threshold_normal_dataset = AD_Dataset(data=threshold_normal.values, config=config)\n",
    "\n",
    "threshold_abnormal_dataset = AD_Dataset(data=threshold_abnormal.values, config=config)\n",
    "test_dataset = AD_Dataset(data=test.values, config=config)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# MODEL"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# model\n",
    "model = LSTMAutoEncoder(config=config)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# TRAIN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 10%|█         | 1/10 [00:01<00:16,  1.80s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "save model\n",
      "epoch : 1, train_loss : 0.5740692019462585, valid_loss : 0.9876536726951599\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 20%|██        | 2/10 [00:03<00:14,  1.83s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "save model\n",
      "epoch : 2, train_loss : 0.42822977900505066, valid_loss : 0.9183894395828247\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 30%|███       | 3/10 [00:05<00:12,  1.79s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch : 3, train_loss : 0.39950976769129437, valid_loss : 0.9502300024032593\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 40%|████      | 4/10 [00:06<00:09,  1.59s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch : 4, train_loss : 0.40024061997731525, valid_loss : 0.9437336325645447\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 50%|█████     | 5/10 [00:07<00:07,  1.46s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch : 5, train_loss : 0.39519686500231427, valid_loss : 0.9976145029067993\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 60%|██████    | 6/10 [00:09<00:05,  1.42s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch : 6, train_loss : 0.3921799858411153, valid_loss : 0.9876083135604858\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 70%|███████   | 7/10 [00:10<00:03,  1.33s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch : 7, train_loss : 0.3904853065808614, valid_loss : 0.9688310623168945\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 80%|████████  | 8/10 [00:11<00:02,  1.30s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 00008: reducing learning rate of group 0 to 2.5000e-02.\n",
      "epoch : 8, train_loss : 0.38819823662439984, valid_loss : 0.9752747416496277\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 90%|█████████ | 9/10 [00:12<00:01,  1.30s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch : 9, train_loss : 0.38731799523035687, valid_loss : 0.9636232852935791\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 10/10 [00:14<00:00,  1.41s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch : 10, train_loss : 0.3862740099430084, valid_loss : 0.9552053213119507\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "# data loader\n",
    "train_loader = DataLoader(train_dataset, batch_size=64, shuffle=False, num_workers=0)\n",
    "valid_loader = DataLoader(valid_dataset, batch_size=64, shuffle=False, num_workers=0)\n",
    "\n",
    "# train\n",
    "best_model, train_loss_arr, valid_loss_arr = trainer(\n",
    "    config=config, model=model, train_loader=train_loader, valid_loader=valid_loader\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.title(\"Train Loss\")\n",
    "plt.plot(train_loss_arr)\n",
    "plt.show()\n",
    "\n",
    "plt.title(\"Validation Loss\")\n",
    "plt.plot(valid_loss_arr)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# ANOMALY SCORE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "threshold_normal_loader = DataLoader(\n",
    "    threshold_normal_dataset, batch_size=64, shuffle=False, num_workers=0\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/kimtaejong/anaconda3/envs/base311/lib/python3.11/site-packages/torch/nn/_reduction.py:42: UserWarning: size_average and reduce args will be deprecated, please use reduction='none' instead.\n",
      "  warnings.warn(warning.format(ret))\n",
      "1it [00:00,  7.62it/s]\n",
      "100%|██████████| 256/256 [00:00<00:00, 40655.10it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "정상구간에서 비정상 점수 분포\n",
      "평균[0.008890953335334304], 중간[0.005415701303390049], 최소[0.00015141544990292266], 최대[0.05381986607703902]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "reconstrcution_error = get_reconstrcution_error(\n",
    "    config, best_model, threshold_normal_loader\n",
    ")\n",
    "\n",
    "mean = reconstrcution_error.mean(axis=0)\n",
    "sigma = np.cov(reconstrcution_error.T)\n",
    "\n",
    "#\n",
    "anomaly_calculator = Anomaly_Calculator(mean=mean, sigma=sigma)\n",
    "score = anomaly_calculator(reconstrcution_error)\n",
    "\n",
    "threshold_anomaly_scores = []\n",
    "\n",
    "for temp_loss in tqdm(reconstrcution_error):\n",
    "    threshold_anomaly_scores.append(anomaly_calculator(temp_loss))\n",
    "\n",
    "# 정상구간에서 비정상 점수 분포\n",
    "print(\"정상구간에서 비정상 점수 분포\")\n",
    "print(\n",
    "    \"평균[{}], 중간[{}], 최소[{}], 최대[{}]\".format(\n",
    "        np.mean(threshold_anomaly_scores),\n",
    "        np.median(threshold_anomaly_scores),\n",
    "        np.min(threshold_anomaly_scores),\n",
    "        np.max(threshold_anomaly_scores),\n",
    "    )\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "threshold = 0.0089"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# abnormal data\n",
    "threshold_abnormal_loader = DataLoader(\n",
    "    threshold_abnormal_dataset, batch_size=64, shuffle=False, num_workers=0\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "reconstrcution_error = get_reconstrcution_error(\n",
    "    config, best_model, threshold_abnormal_loader\n",
    ")\n",
    "\n",
    "mean = reconstrcution_error.mean(axis=0)\n",
    "sigma = np.cov(reconstrcution_error.T)\n",
    "\n",
    "#\n",
    "anomaly_calculator = Anomaly_Calculator(mean=mean, sigma=sigma)\n",
    "score = anomaly_calculator(reconstrcution_error)\n",
    "\n",
    "threshold_anomaly_scores = []\n",
    "\n",
    "for temp_loss in tqdm(reconstrcution_error):\n",
    "    threshold_anomaly_scores.append(anomaly_calculator(temp_loss))\n",
    "\n",
    "# 정상구간에서 비정상 점수 분포\n",
    "print(\"정상구간에서 비정상 점수 분포\")\n",
    "print(\n",
    "    \"평균[{}], 중간[{}], 최소[{}], 최대[{}]\".format(\n",
    "        np.mean(threshold_anomaly_scores),\n",
    "        np.median(threshold_anomaly_scores),\n",
    "        np.min(threshold_anomaly_scores),\n",
    "        np.max(threshold_anomaly_scores),\n",
    "    )\n",
    ")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "base311",
   "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": "3.11.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}