guidance.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Step-by-step Guidence on How to Install and Use ADBench**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Install ADBench"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-07-19T15:56:47.229748Z",
     "start_time": "2023-07-19T15:56:35.675200Z"
    }
   },
   "outputs": [],
   "source": [
    "!pip install adbench\n",
    "!pip install --upgrade adbench"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-07-19T16:06:57.104897Z",
     "start_time": "2023-07-19T16:06:57.092930Z"
    }
   },
   "outputs": [],
   "source": [
    "# download datasets in ADBench from the remote github repo\n",
    "from adbench.myutils import Utils\n",
    "utils = Utils()\n",
    "# we recommend jihulab for China mainland user and github otherwise\n",
    "utils.download_datasets(repo='jihulab')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "# Run ADBench "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-07-17T15:33:13.848925Z",
     "start_time": "2023-07-17T15:33:13.833498Z"
    }
   },
   "source": [
    "## Run ADBench experiments"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-07-19T14:44:03.627289Z",
     "start_time": "2023-07-19T14:26:41.551958Z"
    },
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "from adbench.run import RunPipeline\n",
    "\n",
    "'''\n",
    "Params:\n",
    "suffix: file name suffix;\n",
    "\n",
    "parallel: running either 'unsupervise', 'semi-supervise', or 'supervise' (AD) algorithms,\n",
    "corresponding to the Angle I: Availability of Ground Truth Labels (Supervision);\n",
    "\n",
    "realistic_synthetic_mode: testing on 'local', 'global', 'dependency', and 'cluster' anomalies, \n",
    "corresponding to the Angle II: Types of Anomalies;\n",
    "\n",
    "noise type: evaluating algorithms on 'duplicated_anomalies', 'irrelevant_features' and 'label_contamination',\n",
    "corresponding to the Angle III: Model Robustness with Noisy and Corrupted Data.\n",
    "'''"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# return the results including [params, model_name, metrics, time_fit, time_inference]\n",
    "# besides, results will be automatically saved in the dataframe and ouputted as csv file in adbench/result folder\n",
    "\n",
    "pipeline = RunPipeline(suffix='ADBench', parallel='semi-supervise', realistic_synthetic_mode=None, noise_type=None)\n",
    "results = pipeline.run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "pipeline = RunPipeline(suffix='ADBench', parallel='unsupervise', realistic_synthetic_mode='cluster', noise_type=None)\n",
    "results = pipeline.run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "pipeline = RunPipeline(suffix='ADBench', parallel='supervise', realistic_synthetic_mode=None, noise_type='irrelevant_features')\n",
    "results = pipeline.run()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Run your customized algorithm on ADBench datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# customized model on ADBench's datasets\n",
    "from adbench.run import RunPipeline\n",
    "from adbench.baseline.Customized.run import Customized\n",
    "\n",
    "# notice that you should specify the corresponding category of your customized AD algorithm\n",
    "# for example, here we use Logistic Regression as customized clf, which belongs to the supervised algorithm\n",
    "# for your own algorithm, you can realize the same usage as other baselines by modifying the fit.py, model.py, and run.py files in the adbench/baseline/Customized\n",
    "pipeline = RunPipeline(suffix='ADBench', parallel='supervise', realistic_synthetic_mode=None, noise_type=None)\n",
    "results = pipeline.run(clf=Customized)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Run your customized algorithm on customized dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# customized model on customized dataset\n",
    "import numpy as np\n",
    "dataset = {}\n",
    "dataset['X'] = np.random.randn(1000, 20)\n",
    "dataset['y'] = np.random.choice([0, 1], 1000)\n",
    "results = pipeline.run(dataset=dataset, clf=Customized)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Import AD algorithms from ADBench"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "X_train = np.random.randn(1000, 20)\n",
    "y_train = np.random.choice([0, 1], 1000)\n",
    "X_test = np.random.randn(100, 20)\n",
    "\n",
    "# Directly import AD algorithms from the existing toolkits like PyOD\n",
    "from adbench.baseline.PyOD import PYOD\n",
    "model = PYOD(seed=42, model_name='XGBOD')  # initialization\n",
    "model.fit(X_train, y_train)  # fit\n",
    "score = model.predict_score(X_test)  # predict\n",
    "\n",
    "# Import deep learning AD algorithms from our ADBench\n",
    "from adbench.baseline.PReNet.run import PReNet\n",
    "model = PReNet(seed=42)\n",
    "model.fit(X_train, y_train)  # fit\n",
    "score = model.predict_score(X_test)  # predict"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.7.7"
  },
  "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
}