{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# IsoTree to TreeLite\n",
    "\n",
    "This is a short example of converting an Isolation Forest model generated through the [isotree](https://github.com/david-cortes/isotree) library to [treelite](https://treelite.readthedocs.io/en/latest/index.html) format, which can then be passed to the [tl2cgen](https://tl2cgen.readthedocs.io/en/latest/) library to compile these trees to a standalone runtime library which is oftentimes faster ar making predictions.\n",
    "\n",
    "** *\n",
    "### Getting some medium-size data from scikit-learn to fit a model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(20640, 8)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.datasets import fetch_california_housing\n",
    "\n",
    "X, y = fetch_california_housing(return_X_y=True)\n",
    "print(X.shape)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Fitting an isolation forest model through isotree\n",
    "\n",
    "*Note: only models that use `ndim=1` can be exported to `treelite` format.*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from isotree import IsolationForest\n",
    "\n",
    "iso = IsolationForest(ndim=1, ntrees=100, sample_size=256,\n",
    "                      missing_action=\"impute\", max_depth=8)\n",
    "iso.fit(X)\n",
    "\n",
    "### Now convert\n",
    "treelite_model = iso.to_treelite()\n",
    "\n",
    "### OPTIONAL: add annotations for better branch prediction\n",
    "import tl2cgen\n",
    "tl2cgen.annotate_branch(\n",
    "    model=treelite_model,\n",
    "    dmat=tl2cgen.DMatrix(X),\n",
    "    verbose=False,\n",
    "    path=\"iso_branches_annotation.json\"\n",
    ")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Compiling the treelite model\n",
    "\n",
    "These models need to be compiled into a shared library in order to be used:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%capture\n",
    "import tl2cgen\n",
    "import multiprocessing\n",
    "\n",
    "tl2cgen.export_lib(\n",
    "    model=treelite_model,\n",
    "    toolchain=\"clang\",\n",
    "    libpath='./predictor.so',\n",
    "    params={\n",
    "        \"parallel_comp\": multiprocessing.cpu_count(),\n",
    "        \"annotate_in\": \"iso_branches_annotation.json\"\n",
    "    }\n",
    ")\n",
    "treelite_predictor = tl2cgen.Predictor(\"predictor.so\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now verify that they make the same predictions:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.47006444, 0.47770081, 0.4910637 , 0.42605826, 0.41548625,\n",
       "       0.41730139, 0.41699421, 0.43228664, 0.40877799, 0.41800632])"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "iso.predict(X[:10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.47006445],\n",
       "       [0.47770081],\n",
       "       [0.4910637 ],\n",
       "       [0.42605827],\n",
       "       [0.41548626],\n",
       "       [0.41730139],\n",
       "       [0.41699421],\n",
       "       [0.43228664],\n",
       "       [0.40877799],\n",
       "       [0.41800632]])"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "treelite_predictor.predict(tl2cgen.DMatrix(X[:10]))"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "*Note: some small disagreement between the two is expected due to loss of precision when converting. See the documentation in `isotree` for more details.*"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Comparing prediction times"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "20.4 ms ± 423 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n"
     ]
    }
   ],
   "source": [
    "%%timeit\n",
    "import joblib\n",
    "### see docs for 'IsolationForest.predict' about this part\n",
    "iso.set_params(nthreads=joblib.cpu_count(only_physical_cores=True))\n",
    "iso.predict(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4.14 ms ± 620 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)\n"
     ]
    }
   ],
   "source": [
    "%%timeit\n",
    "treelite_predictor.predict(tl2cgen.DMatrix(X))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "base",
   "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": "3.9.16"
  },
  "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
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}