{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from sklearn.datasets import load_iris\n",
    "from sklearn.preprocessing import LabelBinarizer as skLabelBinarizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "class LabelBinarizer():\n",
    "    def __init__(self, pos_label=1, neg_label=0):\n",
    "        self.pos_label = pos_label\n",
    "        self.neg_label = neg_label\n",
    "\n",
    "    def fit(self, y):\n",
    "        self.classes_ = np.unique(y)\n",
    "        return self\n",
    "\n",
    "    def transform(self, y):\n",
    "        y_enc = np.full((y.shape[0], len(self.classes_)), self.neg_label)\n",
    "        for i, c in enumerate(self.classes_):\n",
    "            y_enc[y == c, i] = self.pos_label\n",
    "        if len(self.classes_) == 2:\n",
    "            y_enc = y_enc[:, 1].reshape(-1, 1)\n",
    "        return y_enc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "iris = load_iris()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# binary\n",
    "y = iris.target\n",
    "y = y[y != 2]\n",
    "le1 = LabelBinarizer().fit(y)\n",
    "le2 = skLabelBinarizer().fit(y)\n",
    "assert np.array_equal(le1.classes_, le2.classes_)\n",
    "yt1 = le1.transform(y)\n",
    "yt2 = le2.transform(y)\n",
    "assert yt1.shape == (len(y), 1)\n",
    "assert np.array_equal(yt1, yt2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# numeric multiclass\n",
    "y = iris.target\n",
    "le1 = LabelBinarizer().fit(y)\n",
    "le2 = skLabelBinarizer().fit(y)\n",
    "assert np.array_equal(le1.classes_, le2.classes_)\n",
    "yt1 = le1.transform(y)\n",
    "yt2 = le2.transform(y)\n",
    "assert np.array_equal(yt1, yt2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# string multiclass\n",
    "y = iris.target_names[iris.target]\n",
    "le1 = LabelBinarizer().fit(y)\n",
    "le2 = skLabelBinarizer().fit(y)\n",
    "assert np.array_equal(le1.classes_, le2.classes_)\n",
    "yt1 = le1.transform(y)\n",
    "yt2 = le2.transform(y)\n",
    "assert np.array_equal(yt1, yt2)"
   ]
  }
 ],
 "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}