notebooks/isolation_forest.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Pipeline for the anomaly detection on the SKAB using Isolation Forest"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Details regarding the SKAB one can find in the [SKAB repository](https://github.com/waico/SKAB)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Libraries importing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# libraries importing\n",
    "import sys\n",
    "import warnings\n",
    "\n",
    "import pandas as pd\n",
    "\n",
    "sys.path.append(\"..\")\n",
    "from core.Isolation_Forest import Isolation_Forest\n",
    "from core.metrics import chp_score\n",
    "from core.utils import load_preprocess_skab, plot_results\n",
    "\n",
    "warnings.filterwarnings(\"ignore\", category=UserWarning)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "Xy_traintest_list = load_preprocess_skab()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Method"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Method initialization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# hyperparameters selection\n",
    "random_state = 0\n",
    "n_jobs = -1\n",
    "contamination = 0.0005\n",
    "PARAMS = [random_state, n_jobs, contamination]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# model defining\n",
    "model = Isolation_Forest(PARAMS)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Method fitting and applying"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# inference\n",
    "predicted_outlier, predicted_cp = [], []\n",
    "true_outlier, true_cp = [], []\n",
    "for X_train, X_test, y_train, y_test in Xy_traintest_list:\n",
    "    # classifier fitting\n",
    "    model.fit(X_train)\n",
    "\n",
    "    # results predicting\n",
    "    prediction = (\n",
    "        pd.Series(\n",
    "            model.predict(X_test) * -1,\n",
    "            index=X_test.index,\n",
    "        )\n",
    "        .rolling(3)\n",
    "        .median()\n",
    "        .fillna(0)\n",
    "        .replace(-1, 0)\n",
    "    )\n",
    "\n",
    "    # predicted outliers saving\n",
    "    predicted_outlier.append(prediction)\n",
    "\n",
    "    # predicted CPs saving\n",
    "    prediction_cp = prediction.rolling(30).max().fillna(0).diff().abs()\n",
    "    prediction_cp[0] = prediction[0]\n",
    "    predicted_cp.append(prediction_cp)\n",
    "\n",
    "    true_outlier.append(y_test[\"anomaly\"])\n",
    "    true_cp.append(y_test[\"changepoint\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Results visualization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# [donotremove]\n",
    "plot_results(\n",
    "    (true_outlier[1], predicted_outlier[1]),\n",
    "    (true_cp[1], predicted_cp[1]),\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluation (metrics calculation)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle\n",
    "\n",
    "pickle.dump(\n",
    "    predicted_outlier,\n",
    "    open(\n",
    "        f\"../results/../results/results-{model.__class__.__name__}.pkl\", \"wb\"\n",
    "    ),\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Binary classification (outlier detection) metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "False Alarm Rate 2.56 %\n",
      "Missing Alarm Rate 82.89 %\n",
      "F1 metric 0.29\n"
     ]
    }
   ],
   "source": [
    "# [donotremove]\n",
    "# binary classification metrics calculation\n",
    "binary = chp_score(true_outlier, predicted_outlier, metric=\"binary\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Changepoint detection metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Amount of true anomalies 127\n",
      "A number of missed CPs = 76\n",
      "A number of FPs = 135\n",
      "Average time 0 days 00:00:22.725490196\n"
     ]
    }
   ],
   "source": [
    "# [donotremove]\n",
    "# average detection delay metric calculation\n",
    "add = chp_score(\n",
    "    true_cp,\n",
    "    predicted_cp,\n",
    "    metric=\"average_time\",\n",
    "    window_width=\"60s\",\n",
    "    anomaly_window_destination=\"righter\",\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Standard  -  26.16\n",
      "LowFP  -  19.5\n",
      "LowFN  -  30.82\n"
     ]
    }
   ],
   "source": [
    "# [donotremove]\n",
    "# nab metric calculation\n",
    "nab = chp_score(\n",
    "    true_cp,\n",
    "    predicted_cp,\n",
    "    metric=\"nab\",\n",
    "    window_width=\"60s\",\n",
    "    anomaly_window_destination=\"righter\",\n",
    ")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "",
   "language": "python",
   "name": ""
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.14"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}