notebooks/MSET.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Pipeline for the anomaly detection on the SKAB using MSET"
   ]
  },
  {
   "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": {
    "tags": []
   },
   "source": [
    "## Libraries importing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# libraries importing\n",
    "import math\n",
    "import sys\n",
    "import warnings\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "sys.path.append(\"..\")\n",
    "from core.MSET import MSET\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": {
    "tags": []
   },
   "source": [
    "### Method initialization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def movmean(array, window):\n",
    "    n = np.size(array)\n",
    "    xx = array.copy()\n",
    "    y = []\n",
    "    for i in range(0, window):\n",
    "        y.append(np.roll(xx.tolist() + [np.nan] * window, i))\n",
    "    y = np.nanmean(y, axis=0)\n",
    "    window_ceil = math.ceil(window / 2)\n",
    "\n",
    "    return y[window_ceil - 1 : n + window_ceil - 1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Method fitting and applying"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# inference\n",
    "predicted_outlier, predicted_cp = [], []\n",
    "true_outlier, true_cp = [], []\n",
    "rel_errors = []\n",
    "x = 0\n",
    "for X_train, X_test, y_train, y_test in Xy_traintest_list:\n",
    "    x += 1\n",
    "\n",
    "    model = MSET()\n",
    "    model.fit(X_train)\n",
    "\n",
    "    # results predicting\n",
    "    Y_pred = model.predict(X_test)\n",
    "    err = np.linalg.norm(X_test - Y_pred, axis=1)\n",
    "    rel_err = movmean(err / np.linalg.norm(Y_pred, axis=1), window=60)\n",
    "    rel_errors.append(rel_err)\n",
    "\n",
    "    UCL = 0.01\n",
    "    prediction = (\n",
    "        pd.DataFrame((rel_err > UCL), X_test.index)\n",
    "        .fillna(0)\n",
    "        .any(axis=1)\n",
    "        .astype(int)\n",
    "    )\n",
    "\n",
    "    # predicted outliers saving\n",
    "    predicted_outlier.append(prediction)\n",
    "\n",
    "    # predicted CPs saving\n",
    "    prediction_cp = abs(prediction.diff())\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(f\"../results/results-{model.__class__.__name__}.pkl\", \"wb\"),\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 39.73 %\n",
      "Missing Alarm Rate 14.13 %\n",
      "F1 metric 0.78\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 = 96\n",
      "A number of FPs = 66\n",
      "Average time 0 days 00:00:33.419354838\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  -  13.84\n",
      "LowFP  -  10.22\n",
      "LowFN  -  17.37\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
}