{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Vamos utilizar o KNN e o train_test_split\n",
    "# Importe os módulos necessários\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "colunas = ['buying',\n",
    "'maint',\n",
    "'doors',\n",
    "'persons',\n",
    "'lug_boot',\n",
    "'safety',\n",
    "'y']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Leia o dataset\n",
    "data = _____('data_car.csv', names=colunas)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Veja os primeiros 5 dados\n",
    "data.____"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Veja se tem dado nulo\n",
    "data.___.___"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Transforme os dados categóricos em númericos\n",
    "y_mapping = __________\n",
    "\n",
    "data[___] = data[___].____"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Crie o conjunto de rótulos e o conjunto de features\n",
    "y = ______\n",
    "\n",
    "X = ______"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Construa o conjunto de treino e teste usando:\n",
    "# - 0,2 para o tamanho do conjunto de testes \n",
    "# - 42 para o estado aleatório\n",
    "# - Estratifique a divisão com o conjunto de rótulos \n",
    "\n",
    "X_train, X_test, y_train, y_test = ______(___, ___, test_size=___, random_state=___, stratify=___)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Gerando um gráfico com a acuracia da resposta do modelo com o conjunto de treino e teste e vendo qual a quantidade de vizinhos da um melhor resultado e quais valores da overfiting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Verifiquem a resposta do gráfico com 100 e depois\n",
    "# façam um gráfico com 30\n",
    "\n",
    "neighbors = np.arange(1, 100)\n",
    "train_accuracy = np.empty(len(neighbors))\n",
    "test_accuracy = np.empty(len(neighbors))\n",
    "\n",
    "# Coloque o numero de vizinhos igual a k do modelo do KNN\n",
    "for i, k in enumerate(neighbors):\n",
    "    knn = _____\n",
    "    knn.fit(___, ___)\n",
    "    train_accuracy[i] = knn.score(___, ___)\n",
    "    test_accuracy[i] = knn.score(___, ___)\n",
    "\n",
    "# gerando o grafico\n",
    "plt.title('k-NN: Número de vizinhos')\n",
    "plt.plot(neighbors, test_accuracy, label = 'Accuracia teste')\n",
    "plt.plot(neighbors, train_accuracy, label = 'Accuracia treino')\n",
    "plt.legend()\n",
    "plt.xlabel('Número de vizinhos')\n",
    "plt.ylabel('Accuracia')\n",
    "plt.show()"
   ]
  }
 ],
 "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": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}