7DaysofCode_Modelo.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "70284cff",
   "metadata": {},
   "source": [
    "<font color = 'green' size =6>7 Days Of Code Dia 2 e 3: Classificação de Popularidade</font>\n",
    "\n",
    "Neste dia vou utilizar os dados tratados do dia 1 para realizar um modelo de classificação de Popularidade.\n",
    "\n",
    "Este modelo se trata de um modelo supervisionado do tipo classificação. \n",
    "\n",
    "Tarefas Neste notebook:\n",
    "1. Decidir critério para decidir popularidade (o menos enviesado possível)\n",
    "2. Analisar equilíbrio dos dados\n",
    "3. Criar modelo básico e Dummy\n",
    "4. Avaliar diferentes métricas\n",
    "5. Comparar com 2 modelos diferentes\n",
    "6. Averificar Hiperparâmetros\n",
    "\n",
    "\n",
    "# Importar Bibliotecas e Ler os Dados\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "5f6aa3a9",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "af4276e3",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>popularity</th>\n",
       "      <th>duration_ms</th>\n",
       "      <th>danceability</th>\n",
       "      <th>energy</th>\n",
       "      <th>key</th>\n",
       "      <th>loudness</th>\n",
       "      <th>mode</th>\n",
       "      <th>speechiness</th>\n",
       "      <th>acousticness</th>\n",
       "      <th>instrumentalness</th>\n",
       "      <th>liveness</th>\n",
       "      <th>valence</th>\n",
       "      <th>tempo</th>\n",
       "      <th>time_signature</th>\n",
       "      <th>explicit</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.73</td>\n",
       "      <td>0.044043</td>\n",
       "      <td>0.686294</td>\n",
       "      <td>0.4610</td>\n",
       "      <td>0.090909</td>\n",
       "      <td>0.791392</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.148187</td>\n",
       "      <td>0.032329</td>\n",
       "      <td>0.000001</td>\n",
       "      <td>0.3580</td>\n",
       "      <td>0.718593</td>\n",
       "      <td>0.361245</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.55</td>\n",
       "      <td>0.028566</td>\n",
       "      <td>0.426396</td>\n",
       "      <td>0.1660</td>\n",
       "      <td>0.090909</td>\n",
       "      <td>0.597377</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.079067</td>\n",
       "      <td>0.927711</td>\n",
       "      <td>0.000006</td>\n",
       "      <td>0.1010</td>\n",
       "      <td>0.268342</td>\n",
       "      <td>0.318397</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.57</td>\n",
       "      <td>0.040255</td>\n",
       "      <td>0.444670</td>\n",
       "      <td>0.3590</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.736123</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.057720</td>\n",
       "      <td>0.210843</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.1170</td>\n",
       "      <td>0.120603</td>\n",
       "      <td>0.313643</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.71</td>\n",
       "      <td>0.038557</td>\n",
       "      <td>0.270051</td>\n",
       "      <td>0.0596</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.573701</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.037617</td>\n",
       "      <td>0.908635</td>\n",
       "      <td>0.000071</td>\n",
       "      <td>0.1320</td>\n",
       "      <td>0.143719</td>\n",
       "      <td>0.746758</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.82</td>\n",
       "      <td>0.037969</td>\n",
       "      <td>0.627411</td>\n",
       "      <td>0.4430</td>\n",
       "      <td>0.181818</td>\n",
       "      <td>0.737103</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.054508</td>\n",
       "      <td>0.470884</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0829</td>\n",
       "      <td>0.167839</td>\n",
       "      <td>0.492863</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   popularity  duration_ms  danceability  energy       key  loudness  mode  \\\n",
       "0        0.73     0.044043      0.686294  0.4610  0.090909  0.791392   0.0   \n",
       "1        0.55     0.028566      0.426396  0.1660  0.090909  0.597377   1.0   \n",
       "2        0.57     0.040255      0.444670  0.3590  0.000000  0.736123   1.0   \n",
       "3        0.71     0.038557      0.270051  0.0596  0.000000  0.573701   1.0   \n",
       "4        0.82     0.037969      0.627411  0.4430  0.181818  0.737103   1.0   \n",
       "\n",
       "   speechiness  acousticness  instrumentalness  liveness   valence     tempo  \\\n",
       "0     0.148187      0.032329          0.000001    0.3580  0.718593  0.361245   \n",
       "1     0.079067      0.927711          0.000006    0.1010  0.268342  0.318397   \n",
       "2     0.057720      0.210843          0.000000    0.1170  0.120603  0.313643   \n",
       "3     0.037617      0.908635          0.000071    0.1320  0.143719  0.746758   \n",
       "4     0.054508      0.470884          0.000000    0.0829  0.167839  0.492863   \n",
       "\n",
       "   time_signature  explicit  \n",
       "0             0.8       0.0  \n",
       "1             0.8       0.0  \n",
       "2             0.8       0.0  \n",
       "3             0.6       0.0  \n",
       "4             0.8       0.0  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('https://raw.githubusercontent.com/KaueAbbe/7DaysofCode_ML/main/dataset_normalizado.csv')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b2bbb1db",
   "metadata": {},
   "source": [
    "# 1. Critério de Popularidade\n",
    "## Como se distribui a popularidade?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "a070a498",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([ 8026.,  7972., 16648., 10719., 13131., 10278.,  4576.,  1829.,\n",
       "          396.,    34.]),\n",
       " array([0. , 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1. ]),\n",
       " <BarContainer object of 10 artists>)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(df.popularity)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c6369c5c",
   "metadata": {},
   "source": [
    "Tenho muito mais valores de dados abaixo de 0.5 (metade de 1) do que acima de 0.5. Caso o critério seja maior que 0.8 para ser considerado popular terá muito pouco popular comparados com impopular. Vou utilizar primeiro como critérios acima de 0.6 pois já é difícil ter valores maior que 0.6. \n",
    "\n",
    "## Aplicando Critério"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "2252f8c2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.73, 0.55, 0.57, ..., 0.22, 0.22, 0.22])"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "valores_popular = df.popularity.values\n",
    "valores_popular "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "bb9c374e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1., 0., 0., ..., 0., 0., 0.])"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "condlist = [valores_popular >= 0.6]\n",
    "choicelist = [valores_popular]\n",
    "clas_popular = np.select(condlist, [np.ones((1,len(choicelist[0])))], 0)\n",
    "clas_popular[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "9fba6837",
   "metadata": {},
   "outputs": [],
   "source": [
    "df['popularityClassifier'] = clas_popular[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "a0a2acc3",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>popularity</th>\n",
       "      <th>duration_ms</th>\n",
       "      <th>danceability</th>\n",
       "      <th>energy</th>\n",
       "      <th>key</th>\n",
       "      <th>loudness</th>\n",
       "      <th>mode</th>\n",
       "      <th>speechiness</th>\n",
       "      <th>acousticness</th>\n",
       "      <th>instrumentalness</th>\n",
       "      <th>liveness</th>\n",
       "      <th>valence</th>\n",
       "      <th>tempo</th>\n",
       "      <th>time_signature</th>\n",
       "      <th>explicit</th>\n",
       "      <th>popularityClassifier</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.73</td>\n",
       "      <td>0.044043</td>\n",
       "      <td>0.686294</td>\n",
       "      <td>0.4610</td>\n",
       "      <td>0.090909</td>\n",
       "      <td>0.791392</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.148187</td>\n",
       "      <td>0.032329</td>\n",
       "      <td>0.000001</td>\n",
       "      <td>0.3580</td>\n",
       "      <td>0.718593</td>\n",
       "      <td>0.361245</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.55</td>\n",
       "      <td>0.028566</td>\n",
       "      <td>0.426396</td>\n",
       "      <td>0.1660</td>\n",
       "      <td>0.090909</td>\n",
       "      <td>0.597377</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.079067</td>\n",
       "      <td>0.927711</td>\n",
       "      <td>0.000006</td>\n",
       "      <td>0.1010</td>\n",
       "      <td>0.268342</td>\n",
       "      <td>0.318397</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.57</td>\n",
       "      <td>0.040255</td>\n",
       "      <td>0.444670</td>\n",
       "      <td>0.3590</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.736123</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.057720</td>\n",
       "      <td>0.210843</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.1170</td>\n",
       "      <td>0.120603</td>\n",
       "      <td>0.313643</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.71</td>\n",
       "      <td>0.038557</td>\n",
       "      <td>0.270051</td>\n",
       "      <td>0.0596</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.573701</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.037617</td>\n",
       "      <td>0.908635</td>\n",
       "      <td>0.000071</td>\n",
       "      <td>0.1320</td>\n",
       "      <td>0.143719</td>\n",
       "      <td>0.746758</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.82</td>\n",
       "      <td>0.037969</td>\n",
       "      <td>0.627411</td>\n",
       "      <td>0.4430</td>\n",
       "      <td>0.181818</td>\n",
       "      <td>0.737103</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.054508</td>\n",
       "      <td>0.470884</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0829</td>\n",
       "      <td>0.167839</td>\n",
       "      <td>0.492863</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   popularity  duration_ms  danceability  energy       key  loudness  mode  \\\n",
       "0        0.73     0.044043      0.686294  0.4610  0.090909  0.791392   0.0   \n",
       "1        0.55     0.028566      0.426396  0.1660  0.090909  0.597377   1.0   \n",
       "2        0.57     0.040255      0.444670  0.3590  0.000000  0.736123   1.0   \n",
       "3        0.71     0.038557      0.270051  0.0596  0.000000  0.573701   1.0   \n",
       "4        0.82     0.037969      0.627411  0.4430  0.181818  0.737103   1.0   \n",
       "\n",
       "   speechiness  acousticness  instrumentalness  liveness   valence     tempo  \\\n",
       "0     0.148187      0.032329          0.000001    0.3580  0.718593  0.361245   \n",
       "1     0.079067      0.927711          0.000006    0.1010  0.268342  0.318397   \n",
       "2     0.057720      0.210843          0.000000    0.1170  0.120603  0.313643   \n",
       "3     0.037617      0.908635          0.000071    0.1320  0.143719  0.746758   \n",
       "4     0.054508      0.470884          0.000000    0.0829  0.167839  0.492863   \n",
       "\n",
       "   time_signature  explicit  popularityClassifier  \n",
       "0             0.8       0.0                   1.0  \n",
       "1             0.8       0.0                   0.0  \n",
       "2             0.8       0.0                   0.0  \n",
       "3             0.6       0.0                   1.0  \n",
       "4             0.8       0.0                   1.0  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d4a0d213",
   "metadata": {},
   "source": [
    "# 2. Balanceando Dados"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "535f365b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0    65984\n",
       "1.0     7625\n",
       "Name: popularityClassifier, dtype: int64"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.popularityClassifier.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "2b564814",
   "metadata": {},
   "outputs": [],
   "source": [
    "from imblearn.over_sampling import RandomOverSampler\n",
    "X = df.drop('popularityClassifier', axis = 1)\n",
    "y = df['popularityClassifier']\n",
    "ros = RandomOverSampler(random_state=0)\n",
    "X_resampled, y_resampled = ros.fit_resample(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "eaf7a769",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>duration_ms</th>\n",
       "      <th>danceability</th>\n",
       "      <th>energy</th>\n",
       "      <th>key</th>\n",
       "      <th>loudness</th>\n",
       "      <th>mode</th>\n",
       "      <th>speechiness</th>\n",
       "      <th>acousticness</th>\n",
       "      <th>instrumentalness</th>\n",
       "      <th>liveness</th>\n",
       "      <th>valence</th>\n",
       "      <th>tempo</th>\n",
       "      <th>time_signature</th>\n",
       "      <th>explicit</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.044043</td>\n",
       "      <td>0.686294</td>\n",
       "      <td>0.4610</td>\n",
       "      <td>0.090909</td>\n",
       "      <td>0.791392</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.148187</td>\n",
       "      <td>0.032329</td>\n",
       "      <td>0.000001</td>\n",
       "      <td>0.3580</td>\n",
       "      <td>0.718593</td>\n",
       "      <td>0.361245</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.028566</td>\n",
       "      <td>0.426396</td>\n",
       "      <td>0.1660</td>\n",
       "      <td>0.090909</td>\n",
       "      <td>0.597377</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.079067</td>\n",
       "      <td>0.927711</td>\n",
       "      <td>0.000006</td>\n",
       "      <td>0.1010</td>\n",
       "      <td>0.268342</td>\n",
       "      <td>0.318397</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.040255</td>\n",
       "      <td>0.444670</td>\n",
       "      <td>0.3590</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.736123</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.057720</td>\n",
       "      <td>0.210843</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.1170</td>\n",
       "      <td>0.120603</td>\n",
       "      <td>0.313643</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.038557</td>\n",
       "      <td>0.270051</td>\n",
       "      <td>0.0596</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.573701</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.037617</td>\n",
       "      <td>0.908635</td>\n",
       "      <td>0.000071</td>\n",
       "      <td>0.1320</td>\n",
       "      <td>0.143719</td>\n",
       "      <td>0.746758</td>\n",
       "      <td>0.6</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.037969</td>\n",
       "      <td>0.627411</td>\n",
       "      <td>0.4430</td>\n",
       "      <td>0.181818</td>\n",
       "      <td>0.737103</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.054508</td>\n",
       "      <td>0.470884</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0829</td>\n",
       "      <td>0.167839</td>\n",
       "      <td>0.492863</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>131963</th>\n",
       "      <td>0.040450</td>\n",
       "      <td>0.394924</td>\n",
       "      <td>0.4310</td>\n",
       "      <td>0.636364</td>\n",
       "      <td>0.767068</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.035648</td>\n",
       "      <td>0.250000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.1300</td>\n",
       "      <td>0.276382</td>\n",
       "      <td>0.584044</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>131964</th>\n",
       "      <td>0.050064</td>\n",
       "      <td>0.663959</td>\n",
       "      <td>0.8300</td>\n",
       "      <td>0.909091</td>\n",
       "      <td>0.827368</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.037927</td>\n",
       "      <td>0.000632</td>\n",
       "      <td>0.000001</td>\n",
       "      <td>0.1210</td>\n",
       "      <td>0.550754</td>\n",
       "      <td>0.513572</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>131965</th>\n",
       "      <td>0.051744</td>\n",
       "      <td>0.727919</td>\n",
       "      <td>0.2980</td>\n",
       "      <td>0.545455</td>\n",
       "      <td>0.745316</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.040829</td>\n",
       "      <td>0.557229</td>\n",
       "      <td>0.002930</td>\n",
       "      <td>0.0997</td>\n",
       "      <td>0.263317</td>\n",
       "      <td>0.476797</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>131966</th>\n",
       "      <td>0.032863</td>\n",
       "      <td>0.753299</td>\n",
       "      <td>0.5210</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.789579</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.045699</td>\n",
       "      <td>0.025301</td>\n",
       "      <td>0.023000</td>\n",
       "      <td>0.5810</td>\n",
       "      <td>0.649246</td>\n",
       "      <td>0.513749</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>131967</th>\n",
       "      <td>0.043947</td>\n",
       "      <td>0.674112</td>\n",
       "      <td>0.7920</td>\n",
       "      <td>0.545455</td>\n",
       "      <td>0.818249</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.029326</td>\n",
       "      <td>0.184739</td>\n",
       "      <td>0.000417</td>\n",
       "      <td>0.1060</td>\n",
       "      <td>0.864322</td>\n",
       "      <td>0.575091</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>131968 rows × 14 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        duration_ms  danceability  energy       key  loudness  mode  \\\n",
       "0          0.044043      0.686294  0.4610  0.090909  0.791392   0.0   \n",
       "1          0.028566      0.426396  0.1660  0.090909  0.597377   1.0   \n",
       "2          0.040255      0.444670  0.3590  0.000000  0.736123   1.0   \n",
       "3          0.038557      0.270051  0.0596  0.000000  0.573701   1.0   \n",
       "4          0.037969      0.627411  0.4430  0.181818  0.737103   1.0   \n",
       "...             ...           ...     ...       ...       ...   ...   \n",
       "131963     0.040450      0.394924  0.4310  0.636364  0.767068   1.0   \n",
       "131964     0.050064      0.663959  0.8300  0.909091  0.827368   1.0   \n",
       "131965     0.051744      0.727919  0.2980  0.545455  0.745316   0.0   \n",
       "131966     0.032863      0.753299  0.5210  0.000000  0.789579   1.0   \n",
       "131967     0.043947      0.674112  0.7920  0.545455  0.818249   1.0   \n",
       "\n",
       "        speechiness  acousticness  instrumentalness  liveness   valence  \\\n",
       "0          0.148187      0.032329          0.000001    0.3580  0.718593   \n",
       "1          0.079067      0.927711          0.000006    0.1010  0.268342   \n",
       "2          0.057720      0.210843          0.000000    0.1170  0.120603   \n",
       "3          0.037617      0.908635          0.000071    0.1320  0.143719   \n",
       "4          0.054508      0.470884          0.000000    0.0829  0.167839   \n",
       "...             ...           ...               ...       ...       ...   \n",
       "131963     0.035648      0.250000          0.000000    0.1300  0.276382   \n",
       "131964     0.037927      0.000632          0.000001    0.1210  0.550754   \n",
       "131965     0.040829      0.557229          0.002930    0.0997  0.263317   \n",
       "131966     0.045699      0.025301          0.023000    0.5810  0.649246   \n",
       "131967     0.029326      0.184739          0.000417    0.1060  0.864322   \n",
       "\n",
       "           tempo  time_signature  explicit  \n",
       "0       0.361245             0.8       0.0  \n",
       "1       0.318397             0.8       0.0  \n",
       "2       0.313643             0.8       0.0  \n",
       "3       0.746758             0.6       0.0  \n",
       "4       0.492863             0.8       0.0  \n",
       "...          ...             ...       ...  \n",
       "131963  0.584044             0.8       0.0  \n",
       "131964  0.513572             0.8       0.0  \n",
       "131965  0.476797             0.8       0.0  \n",
       "131966  0.513749             0.8       0.0  \n",
       "131967  0.575091             0.8       0.0  \n",
       "\n",
       "[131968 rows x 14 columns]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_resampled = X_resampled.drop('popularity', axis = 1)\n",
    "X_resampled"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "4aefc16d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.0    65984\n",
       "0.0    65984\n",
       "Name: popularityClassifier, dtype: int64"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_resampled.value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9c286ea4",
   "metadata": {},
   "source": [
    "# 3. Modelo Básico e Dummy\n",
    "**Aqui inicio a fase de teste e otimização**\n",
    "## Dummy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "4a5401ec",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.dummy import DummyClassifier\n",
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "b73bab70",
   "metadata": {},
   "outputs": [],
   "source": [
    "X, x_validacao, Y, y_validacao = train_test_split(X_resampled, y_resampled, test_size=0.2, shuffle=True)\n",
    "x_treino, x_teste, y_treino, y_teste = train_test_split(X, Y, test_size=0.2, shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "a142de18",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "DummyClassifier()"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Dummy = DummyClassifier(strategy='prior')\n",
    "Dummy.fit(x_treino, y_treino)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "eaf2ae57",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.5"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Dummy.score(X_resampled, y_resampled)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "adf5fbb7",
   "metadata": {},
   "source": [
    "## Modelo Simples com Cross_validate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "cc7ec062",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import cross_validate\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.model_selection import KFold"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "fb2bebf2",
   "metadata": {},
   "outputs": [],
   "source": [
    "def imprime_resultados(results):\n",
    "  media = results['test_score'].mean()\n",
    "  std = results['test_score'].std()\n",
    "  print('acuracia media: [%.2f]'%( media*100))\n",
    "  print('acuracia intervalo: [%.2f, %.2f]'%( (media-2*std)*100, (media+2*std)*100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "7ad13ecf",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acuracia media: [62.52]\n",
      "acuracia intervalo: [61.15, 63.90]\n"
     ]
    }
   ],
   "source": [
    "SEED = 42\n",
    "np.random.seed(SEED)\n",
    "\n",
    "cv = KFold(n_splits = 10, shuffle = True)\n",
    "\n",
    "modelo = DecisionTreeClassifier(max_depth = 5)\n",
    "results = cross_validate(modelo, x_treino, y_treino, cv = cv)\n",
    "imprime_resultados(results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c24f977d",
   "metadata": {},
   "source": [
    "O modelo do DecisionTree é melhor que o Dummy. Que é um avanço!\n",
    "\n",
    "# 4. Avaliando Métricas\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "54b62439",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import accuracy_score, precision_score,confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "4e06c521",
   "metadata": {},
   "outputs": [],
   "source": [
    "def metricas(yprevisto):\n",
    "    acuracy = round(accuracy_score(y_teste, yprevisto),2)*100\n",
    "    precisao = round(precision_score(y_teste, yprevisto),2)*100\n",
    "    \n",
    "    print(f'Acurácia: {acuracy} \\n Precisão: {precisao}')\n",
    "    \n",
    "    print(confusion_matrix(y_teste,yprevisto))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "0ae4f4c2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Acurácia: 62.0 \n",
      " Precisão: 57.99999999999999\n",
      "[[4001 6575]\n",
      " [1412 9127]]\n"
     ]
    }
   ],
   "source": [
    "modelo.fit(x_treino, y_treino)\n",
    "DT_previsao = modelo.predict(x_teste)\n",
    "metricas(DT_previsao)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6f1f58f0",
   "metadata": {},
   "source": [
    "# 5. Criando outros modelos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "c1019b28",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acuracia media: [55.57]\n",
      "acuracia intervalo: [54.62, 56.52]\n",
      "Acurácia: 55.00000000000001 \n",
      " Precisão: 56.00000000000001\n",
      "[[6766 3810]\n",
      " [5591 4948]]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.naive_bayes import BernoulliNB\n",
    "\n",
    "Bnb = BernoulliNB()\n",
    "results = cross_validate(Bnb, x_treino, y_treino, cv = cv)\n",
    "imprime_resultados(results)\n",
    "Bnb.fit(x_treino, y_treino)\n",
    "Bnbprevisao = Bnb.predict(x_teste)\n",
    "metricas(Bnbprevisao)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "3c2d1f9c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acuracia media: [79.82]\n",
      "acuracia intervalo: [78.76, 80.88]\n",
      "Acurácia: 81.0 \n",
      " Precisão: 74.0\n",
      "[[ 7045  3531]\n",
      " [  403 10136]]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "\n",
    "Knc = KNeighborsClassifier()\n",
    "results = cross_validate(Knc, x_treino, y_treino, cv = cv)\n",
    "imprime_resultados(results)\n",
    "\n",
    "Knc.fit(x_treino, y_treino)\n",
    "Kncprevisao = Knc.predict(x_teste)\n",
    "metricas(Kncprevisao)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1fa4111a",
   "metadata": {},
   "source": [
    "O modelo KneighborsClassifier está tendo resultados melhores que os outros modelos. Na matrix confusão ele apresenta menos erros. Enquanto a árvore de decisão erra dizendo que tem mais negativos que positivos, o Bernoulli diz que tem mais positivos. O Kneighbors apresenta menos erros nos falsos positivos e negativos que os outros dois modelos. \n",
    "\n",
    "Para seguir para os hiperparâmetros utilizarei o KneighborsClassifier\n",
    "\n",
    "# 6. Otimização de Hiperparâmetros"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "id": "61521e06",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3 Vizinhos:\n",
      "acuracia media: [85.29]\n",
      "acuracia intervalo: [84.61, 85.98]\n",
      "4 Vizinhos:\n",
      "acuracia media: [83.76]\n",
      "acuracia intervalo: [82.99, 84.53]\n",
      "5 Vizinhos:\n",
      "acuracia media: [79.64]\n",
      "acuracia intervalo: [78.95, 80.34]\n",
      "6 Vizinhos:\n",
      "acuracia media: [77.95]\n",
      "acuracia intervalo: [77.45, 78.45]\n",
      "7 Vizinhos:\n",
      "acuracia media: [75.40]\n",
      "acuracia intervalo: [74.31, 76.48]\n",
      "8 Vizinhos:\n",
      "acuracia media: [74.38]\n",
      "acuracia intervalo: [73.57, 75.19]\n",
      "9 Vizinhos:\n",
      "acuracia media: [73.11]\n",
      "acuracia intervalo: [72.06, 74.17]\n"
     ]
    }
   ],
   "source": [
    "for i in range(3, 10):\n",
    "    Knc = KNeighborsClassifier(n_neighbors= i)\n",
    "    results = cross_validate(Knc, x_treino, y_treino, cv = cv)\n",
    "    print(f'{i} Vizinhos:')\n",
    "    imprime_resultados(results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "id": "859caf42",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "15 Vizinhos:\n",
      "acuracia media: [70.26]\n",
      "acuracia intervalo: [69.46, 71.05]\n",
      "16 Vizinhos:\n",
      "acuracia media: [70.35]\n",
      "acuracia intervalo: [69.34, 71.35]\n",
      "17 Vizinhos:\n",
      "acuracia media: [69.99]\n",
      "acuracia intervalo: [68.92, 71.06]\n",
      "18 Vizinhos:\n",
      "acuracia media: [69.93]\n",
      "acuracia intervalo: [68.71, 71.14]\n",
      "19 Vizinhos:\n",
      "acuracia media: [69.38]\n",
      "acuracia intervalo: [68.42, 70.34]\n",
      "20 Vizinhos:\n",
      "acuracia media: [69.30]\n",
      "acuracia intervalo: [68.46, 70.13]\n"
     ]
    }
   ],
   "source": [
    "for i in range(15, 21):\n",
    "    Knc = KNeighborsClassifier(n_neighbors= i)\n",
    "    results = cross_validate(Knc, x_treino, y_treino, cv = cv)\n",
    "    print(f'{i} Vizinhos:')\n",
    "    imprime_resultados(results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "id": "98f4a401",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "uniform weight:\n",
      "acuracia media: [79.75]\n",
      "acuracia intervalo: [79.00, 80.49]\n",
      "distance weight:\n",
      "acuracia media: [82.52]\n",
      "acuracia intervalo: [81.42, 83.63]\n"
     ]
    }
   ],
   "source": [
    "for i in ['uniform','distance']:\n",
    "    Knc = KNeighborsClassifier(weights= i)\n",
    "    results = cross_validate(Knc, x_treino, y_treino, cv = cv)\n",
    "    print(f'{i} weight:')\n",
    "    imprime_resultados(results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "id": "ea43517c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2 Vizinhos e uniform weights:\n",
      "acuracia media: [90.86]\n",
      "acuracia intervalo: [90.42, 91.30]\n",
      "2 Vizinhos e distance weights:\n",
      "acuracia media: [91.87]\n",
      "acuracia intervalo: [91.35, 92.39]\n",
      "3 Vizinhos e uniform weights:\n",
      "acuracia media: [85.27]\n",
      "acuracia intervalo: [84.47, 86.06]\n",
      "3 Vizinhos e distance weights:\n",
      "acuracia media: [86.36]\n",
      "acuracia intervalo: [85.63, 87.09]\n",
      "4 Vizinhos e uniform weights:\n",
      "acuracia media: [83.63]\n",
      "acuracia intervalo: [82.57, 84.69]\n",
      "4 Vizinhos e distance weights:\n",
      "acuracia media: [86.53]\n",
      "acuracia intervalo: [86.10, 86.97]\n",
      "5 Vizinhos e uniform weights:\n",
      "acuracia media: [79.74]\n",
      "acuracia intervalo: [78.74, 80.74]\n",
      "5 Vizinhos e distance weights:\n",
      "acuracia media: [82.47]\n",
      "acuracia intervalo: [81.86, 83.08]\n",
      "6 Vizinhos e uniform weights:\n",
      "acuracia media: [77.96]\n",
      "acuracia intervalo: [76.63, 79.30]\n",
      "6 Vizinhos e distance weights:\n",
      "acuracia media: [83.12]\n",
      "acuracia intervalo: [82.72, 83.52]\n"
     ]
    }
   ],
   "source": [
    "for i in range(2, 7):\n",
    "    for y in ['uniform','distance']:\n",
    "        Knc = KNeighborsClassifier(n_neighbors= i, weights = y)\n",
    "        results = cross_validate(Knc, x_treino, y_treino, cv = cv)\n",
    "        print(f'{i} Vizinhos e {y} weights:')\n",
    "        imprime_resultados(results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "id": "7b6fc45e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2 Vizinhos e distance weights:\n",
      "acuracia media: [91.83]\n",
      "acuracia intervalo: [91.29, 92.37]\n",
      "3 Vizinhos e distance weights:\n",
      "acuracia media: [86.40]\n",
      "acuracia intervalo: [85.60, 87.19]\n"
     ]
    }
   ],
   "source": [
    "for i in range(2, 4):\n",
    "    Knc = KNeighborsClassifier(n_neighbors= i, weights = 'distance', p=2)\n",
    "    results = cross_validate(Knc, x_treino, y_treino, cv = cv)\n",
    "    print(f'{i} Vizinhos e {y} weights:')\n",
    "    imprime_resultados(results)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "516ade05",
   "metadata": {},
   "source": [
    "Mexendo em alguns hiperparâmetors vi que os melhores são: n_neighbors = 2, weights = 'distancia, p=2\n",
    "# 7. Validação"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "0699761d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acuracia media: [91.87]\n",
      "acuracia intervalo: [91.43, 92.31]\n",
      "Acurácia: 93.0 \n",
      " Precisão: 87.0\n",
      "[[ 9057  1519]\n",
      " [   47 10492]]\n"
     ]
    }
   ],
   "source": [
    "Knc = KNeighborsClassifier(n_neighbors= 2, weights = 'distance', p=2)\n",
    "results = cross_validate(Knc, x_treino, y_treino, cv = cv)\n",
    "imprime_resultados(results)\n",
    "\n",
    "Knc.fit(x_treino, y_treino)\n",
    "Kncprevisao = Knc.predict(x_teste)\n",
    "metricas(Kncprevisao)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "d86b878e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Acurácia: 93.0 \n",
      " Precisão: 88.0\n"
     ]
    }
   ],
   "source": [
    "Kncvalidacao = Knc.predict(x_validacao)\n",
    "\n",
    "acuracy = round(accuracy_score(y_validacao, Kncvalidacao),2)*100\n",
    "precisao = round(precision_score(y_validacao, Kncvalidacao),2)*100\n",
    "    \n",
    "print(f'Acurácia: {acuracy} \\n Precisão: {precisao}')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "45ade614",
   "metadata": {},
   "source": [
    "O modelo está validado!\n",
    "\n",
    "Com Acurácia Média de 91.79%, e intervalo entre 91.22 e 92.36. Temos como resultados da validação acurácia de 92%, dentro do intervalo. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "id": "68213918",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.45, 91.5, 'Acurácia Média: 91.8% \\nIntervalo da Acurácia: 91.22% e 92.36%.')"
      ]
     },
     "execution_count": 161,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize = (10, 6))\n",
    "plt.bar(x = ['Acurácia','Precisão'], height = [acuracy, precisao], color = ['darkblue','gray'])\n",
    "plt.title('Acurácia e Precisao da Validação', fontsize = 18)\n",
    "plt.ylabel('Porcentagem', fontsize = 14)\n",
    "plt.ylim(0, 100)\n",
    "\n",
    "plt.text(x = 0.45, y = 91.5, s = 'Acurácia Média: 91.8% \\nIntervalo da Acurácia: 91.22% e 92.36%.', fontsize = 12,\n",
    "         fontweight = 'roman')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "af49ec9d",
   "metadata": {},
   "source": [
    "<font color = 'green' size =6>7 Days Of Code Dia 4: Outras Formas de Progredir em Baseline</font>\n",
    "\n",
    "O objetivo neste dia 4 adiante é realizar as tarefas dos dias de forma diferente que eu fiz a primeira vez.\n",
    "Neste dia em específico vou realizar a criação da **baseline**, a qual na primeira vez foi realizada usando DummyClassifier. A baseline Dummy teve 50% de acurácia. \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "ab1bd33a",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "df = pd.read_csv('https://raw.githubusercontent.com/KaueAbbe/7DaysofCode_ML/main/dataset_normalizado.csv')\n",
    "valores_popular = df.popularity.values\n",
    "condlist = [valores_popular >= 0.6]\n",
    "choicelist = [valores_popular]\n",
    "clas_popular = np.select(condlist, [np.ones((1,len(choicelist[0])))], 0)\n",
    "df['popularityClassifier'] = clas_popular[0]\n",
    "\n",
    "from imblearn.over_sampling import RandomOverSampler\n",
    "X = df.drop(['popularityClassifier','duration_ms'], axis = 1)\n",
    "y = df['popularityClassifier']\n",
    "ros = RandomOverSampler(random_state=0)\n",
    "X_resampled, y_resampled = ros.fit_resample(X, y)\n",
    "X_resampled = X_resampled.drop('popularity', axis = 1)\n",
    "\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "X, x_validacao, Y, y_validacao = train_test_split(X_resampled, y_resampled, test_size=0.2, shuffle=True)\n",
    "x_treino, x_teste, y_treino, y_teste = train_test_split(X, Y, test_size=0.2, shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "97650321",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.model_selection import cross_validate\n",
    "from sklearn.model_selection import KFold"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "053fbb27",
   "metadata": {},
   "outputs": [],
   "source": [
    "LR = LogisticRegression()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "4fdc7ffc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LogisticRegression()"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "LR.fit(x_treino, y_treino)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "8f4e4e3d",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_previsto = LR.predict(x_teste)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "1087f342",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acuracia media: [60.95]\n",
      "acuracia intervalo: [60.27, 61.63]\n"
     ]
    }
   ],
   "source": [
    "SEED = 42\n",
    "np.random.seed(SEED)\n",
    "\n",
    "cv = KFold(n_splits = 10, shuffle = True)\n",
    "\n",
    "modelo = LogisticRegression()\n",
    "results = cross_validate(modelo, x_treino, y_treino, cv = cv)\n",
    "imprime_resultados(results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "3024636b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Acurácia: 61.0 \n",
      " Precisão: 59.0\n",
      "[[5543 5074]\n",
      " [3134 7364]]\n"
     ]
    }
   ],
   "source": [
    "metricas(y_previsto)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "e91d621f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 0.58637173, -0.84705047,  0.0070411 , -0.52732455, -0.1124284 ,\n",
       "        -1.45679308, -0.77129392, -1.49759267, -1.50336961, -0.2695139 ,\n",
       "        -0.62652396,  0.35190018,  0.60511139]])"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "LR.coef_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "1d0b7da8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1.64711833])"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "LR.intercept_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "94d1f654",
   "metadata": {},
   "outputs": [],
   "source": [
    "from yellowbrick.model_selection import FeatureImportances"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "0848fa7f",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\kauea\\anaconda3\\lib\\site-packages\\sklearn\\base.py:450: UserWarning: X does not have valid feature names, but LogisticRegression was fitted with feature names\n",
      "  warnings.warn(\n",
      "C:\\Users\\kauea\\anaconda3\\lib\\site-packages\\yellowbrick\\model_selection\\importances.py:194: YellowbrickWarning: detected multi-dimensional feature importances but stack=False, using mean to aggregate them.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x396 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:title={'center':'Feature Importances of 13 Features using LogisticRegression'}, xlabel='relative coefficient magnitude'>"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "viz = FeatureImportances(LR)\n",
    "viz.fit(x_treino, y_treino)\n",
    "viz.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "89ba7d8a",
   "metadata": {},
   "source": [
    "### Anotações\n",
    "Temos então que a mudança de BaseLine para LogisticRegression aumentou o valor da Baseline de **50% para 61.3%**. Neste cenário novo o modelo Bernoulli ficaria abaixo da Baseline, e a DecisionTree ficaria próximo dela. Neste cenário ambos os modelos seriam descartados mais rápido por não apresentar nem 10% de superioridade em relação a base.\n",
    "\n",
    "**Aprendizado para o futuro:** O Dummy é um baseline muito simples, que se for fitado com dados balanceados terá sempre acurácia de 50%. Utilize outro modelo como Baseline para estudar melhor o problema e buscar resultados melhores durante sua pesquisa.\n",
    "\n",
    "<font color = 'green' size =6>7 Days Of Code Dia 5: Outras Formas de Progredir em Métricas</font>\n",
    "\n",
    "Neste dia tenho que realizar a escolha da **métrica**, a qual na primeira vez foi escolhida a acurácia. \n",
    "\n",
    "**Acurácia:** Avalia a quantidade de acertos no geral.\n",
    "\n",
    "**Precisão:** Avalia quantos acertos com classificação positiva levando em consideração os falsos positivos. *Usado em modelos que precisa diminuir os falsos positivos*\n",
    "\n",
    "**Recall:** Avalia quantos acertos com classificação positiva levando em consideração os falsos negativos. *Usado em modelos que precisa diminuir falsos negativos*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "57261790",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import accuracy_score, precision_score,confusion_matrix, recall_score,ConfusionMatrixDisplay\n",
    "disp = ConfusionMatrixDisplay.from_estimator(\n",
    "        Knc,\n",
    "        x_teste,\n",
    "        y_teste,\n",
    "        cmap=plt.cm.Blues)\n",
    "disp.ax_.set_title('Matrix')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "0dad1f71",
   "metadata": {},
   "outputs": [],
   "source": [
    "def metricas_att(yprevisto):\n",
    "    acuracy = round(accuracy_score(y_teste, yprevisto),2)*100\n",
    "    precisao = round(precision_score(y_teste, yprevisto),2)*100\n",
    "    recall =  round(recall_score(y_teste,yprevisto),2)*100\n",
    "    \n",
    "    print(f'Acurácia: {acuracy} \\n Precisão: {precisao} \\n Recall:{recall}')\n",
    "    \n",
    "    disp = ConfusionMatrixDisplay.from_estimator(\n",
    "        Knc,\n",
    "        x_teste,\n",
    "        y_teste,\n",
    "        cmap=plt.cm.Blues)\n",
    "    disp.ax_.set_title('Matrix')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "2442bdec",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acuracia media: [55.53]\n",
      "acuracia intervalo: [54.58, 56.48]\n",
      "Acurácia: 56.00000000000001 \n",
      " Precisão: 56.99999999999999 \n",
      " Recall:47.0\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "Bnb = BernoulliNB()\n",
    "results = cross_validate(Bnb, x_treino, y_treino, cv = cv)\n",
    "imprime_resultados(results)\n",
    "Bnb.fit(x_treino, y_treino)\n",
    "Bnbprevisao = Bnb.predict(x_teste)\n",
    "metricas_att(Bnbprevisao)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "accd36ed",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acuracia media: [79.94]\n",
      "acuracia intervalo: [79.30, 80.59]\n",
      "Acurácia: 81.0 \n",
      " Precisão: 74.0 \n",
      " Recall:96.0\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "Knc = KNeighborsClassifier()\n",
    "results = cross_validate(Knc, x_treino, y_treino, cv = cv)\n",
    "imprime_resultados(results)\n",
    "\n",
    "Knc.fit(x_treino, y_treino)\n",
    "Kncprevisao = Knc.predict(x_teste)\n",
    "metricas_att(Kncprevisao)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1c841f4e",
   "metadata": {},
   "source": [
    "### Anotações\n",
    "No fim optei pela mudança de métrica, de **acurácia para Recall**. O motivo se deve ao fato do Recall levar em consideração os *Falsos Negativos*, e para o contexto de avaliar popularidade de música seria mais interessante minimizar os erros de falso negativos. Dessa forma terá mais acertos quando dizer que a música não será popular.\n",
    "\n",
    "**Aprendizado para o futuro:** Dependendo do problema a qual o modelo foi proposto serão utilizadas outras métricas. Cada métrica tem uma função específica, assim como equação, e por isso tem interpretação diferente. As quantidades de Verdadeiro Positivo, Verdadeiro Negativo, Falso Positivo e Falso Negativo são vistas na Matriz Confusão, e esses valores são utilizados no cálculo de precisão e Recall. \n",
    "\n",
    "<font color = 'green' size =6>7 Days Of Code Dia 6: Outras Formas de Progredir em Balanceamento e Hiperparâmetros</font>\n",
    "\n",
    "## Balanceamento"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "4c4865fb",
   "metadata": {},
   "outputs": [],
   "source": [
    "from imblearn.under_sampling import RandomUnderSampler\n",
    "Xu = df.drop('popularityClassifier', axis = 1)\n",
    "yu = df['popularityClassifier']\n",
    "rus = RandomUnderSampler(random_state=0)\n",
    "Xu_resampled, yu_resampled = rus.fit_resample(Xu, yu)\n",
    "\n",
    "Xu_resampled = Xu_resampled.drop('popularity', axis = 1)\n",
    "X, x_validacao, Y, y_validacao = train_test_split(X_resampled, y_resampled, test_size=0.2, shuffle=True)\n",
    "x_treino, x_teste, y_treino, y_teste = train_test_split(X, Y, test_size=0.2, shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "4f53ab15",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acuracia media: [63.60]\n",
      "acuracia intervalo: [62.58, 64.63]\n"
     ]
    }
   ],
   "source": [
    "SEED = 42\n",
    "np.random.seed(SEED)\n",
    "\n",
    "cv = KFold(n_splits = 10, shuffle = True)\n",
    "\n",
    "modelo = DecisionTreeClassifier(max_depth = 5)\n",
    "results = cross_validate(modelo, x_treino, y_treino, cv = cv)\n",
    "imprime_resultados(results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "aaf156b2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acuracia media: [55.82]\n",
      "acuracia intervalo: [53.65, 57.99]\n",
      "Acurácia: 55.00000000000001 \n",
      " Precisão: 56.99999999999999\n",
      "[[775 442]\n",
      " [646 577]]\n"
     ]
    }
   ],
   "source": [
    "results = cross_validate(Bnb, x_treino, y_treino, cv = cv)\n",
    "imprime_resultados(results)\n",
    "Bnb.fit(x_treino, y_treino)\n",
    "Bnbprevisao = Bnb.predict(x_teste)\n",
    "metricas(Bnbprevisao)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "24ca7d8d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acuracia media: [59.59]\n",
      "acuracia intervalo: [56.94, 62.24]\n",
      "Acurácia: 61.0 \n",
      " Precisão: 60.0\n",
      "[[690 527]\n",
      " [429 794]]\n"
     ]
    }
   ],
   "source": [
    "results = cross_validate(Knc, x_treino, y_treino, cv = cv)\n",
    "imprime_resultados(results)\n",
    "\n",
    "Knc.fit(x_treino, y_treino)\n",
    "Kncprevisao = Knc.predict(x_teste)\n",
    "metricas(Kncprevisao)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7abfc93d",
   "metadata": {},
   "source": [
    "### Anotações\n",
    "A realização do *UnderSampling* diminuiu os resultados das métricas, mostrando que o balanceamento feito por baixo (retirando dados do dataset) não contribuiu com uma melhor aprendizagem. Por este motivo utilizar o *OverSampling* é o método indicado.\n",
    "\n",
    "**Aprendizado para o futuro:** Utilize sempre duas formas de balanceamento, pois a depender do problema pode ser que o método deo UnderSampling traga melhores resultados. Contudo, dê preferências pelo OverSampling já que ele garante mais dados para o fit.\n",
    "\n",
    "## Hiperparâmetros"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "32c50d8e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acuracia media: [79.87]\n",
      "acuracia intervalo: [78.94, 80.80]\n"
     ]
    }
   ],
   "source": [
    "from imblearn.over_sampling import RandomOverSampler\n",
    "X = df.drop('popularityClassifier', axis = 1)\n",
    "y = df['popularityClassifier']\n",
    "ros = RandomOverSampler(random_state=0)\n",
    "X_resampled, y_resampled = ros.fit_resample(X, y)\n",
    "X_resampled = X_resampled.drop('popularity', axis = 1)\n",
    "X, x_validacao, Y, y_validacao = train_test_split(X_resampled, y_resampled, test_size=0.2, shuffle=True)\n",
    "x_treino, x_teste, y_treino, y_teste = train_test_split(X, Y, test_size=0.2, shuffle=True)\n",
    "\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "\n",
    "Knc = KNeighborsClassifier()\n",
    "results = cross_validate(Knc, x_treino, y_treino, cv = cv)\n",
    "imprime_resultados(results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "97a412d6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>mean_fit_time</th>\n",
       "      <th>std_fit_time</th>\n",
       "      <th>mean_score_time</th>\n",
       "      <th>std_score_time</th>\n",
       "      <th>param_n_neighbors</th>\n",
       "      <th>param_p</th>\n",
       "      <th>param_weights</th>\n",
       "      <th>params</th>\n",
       "      <th>split0_test_score</th>\n",
       "      <th>split1_test_score</th>\n",
       "      <th>...</th>\n",
       "      <th>split3_test_score</th>\n",
       "      <th>split4_test_score</th>\n",
       "      <th>split5_test_score</th>\n",
       "      <th>split6_test_score</th>\n",
       "      <th>split7_test_score</th>\n",
       "      <th>split8_test_score</th>\n",
       "      <th>split9_test_score</th>\n",
       "      <th>mean_test_score</th>\n",
       "      <th>std_test_score</th>\n",
       "      <th>rank_test_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.231523</td>\n",
       "      <td>0.036805</td>\n",
       "      <td>3.493722</td>\n",
       "      <td>0.971770</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>uniform</td>\n",
       "      <td>{'n_neighbors': 3, 'p': 1, 'weights': 'uniform'}</td>\n",
       "      <td>0.850343</td>\n",
       "      <td>0.855079</td>\n",
       "      <td>...</td>\n",
       "      <td>0.849870</td>\n",
       "      <td>0.846199</td>\n",
       "      <td>0.853067</td>\n",
       "      <td>0.854724</td>\n",
       "      <td>0.852948</td>\n",
       "      <td>0.857684</td>\n",
       "      <td>0.850562</td>\n",
       "      <td>0.852402</td>\n",
       "      <td>0.003082</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.286101</td>\n",
       "      <td>0.051702</td>\n",
       "      <td>4.142222</td>\n",
       "      <td>1.026447</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>distance</td>\n",
       "      <td>{'n_neighbors': 3, 'p': 1, 'weights': 'distance'}</td>\n",
       "      <td>0.860289</td>\n",
       "      <td>0.864906</td>\n",
       "      <td>...</td>\n",
       "      <td>0.861236</td>\n",
       "      <td>0.855553</td>\n",
       "      <td>0.863012</td>\n",
       "      <td>0.868932</td>\n",
       "      <td>0.862420</td>\n",
       "      <td>0.867630</td>\n",
       "      <td>0.859207</td>\n",
       "      <td>0.862738</td>\n",
       "      <td>0.003764</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.237938</td>\n",
       "      <td>0.026114</td>\n",
       "      <td>2.596392</td>\n",
       "      <td>0.446166</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>uniform</td>\n",
       "      <td>{'n_neighbors': 3, 'p': 2, 'weights': 'uniform'}</td>\n",
       "      <td>0.853422</td>\n",
       "      <td>0.856855</td>\n",
       "      <td>...</td>\n",
       "      <td>0.850343</td>\n",
       "      <td>0.850225</td>\n",
       "      <td>0.852119</td>\n",
       "      <td>0.852475</td>\n",
       "      <td>0.850699</td>\n",
       "      <td>0.857447</td>\n",
       "      <td>0.852694</td>\n",
       "      <td>0.853325</td>\n",
       "      <td>0.002660</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.235935</td>\n",
       "      <td>0.039246</td>\n",
       "      <td>1.974733</td>\n",
       "      <td>0.192140</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>distance</td>\n",
       "      <td>{'n_neighbors': 3, 'p': 2, 'weights': 'distance'}</td>\n",
       "      <td>0.862775</td>\n",
       "      <td>0.866682</td>\n",
       "      <td>...</td>\n",
       "      <td>0.861710</td>\n",
       "      <td>0.859578</td>\n",
       "      <td>0.861354</td>\n",
       "      <td>0.866682</td>\n",
       "      <td>0.860052</td>\n",
       "      <td>0.867630</td>\n",
       "      <td>0.861812</td>\n",
       "      <td>0.863673</td>\n",
       "      <td>0.003163</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.224505</td>\n",
       "      <td>0.032128</td>\n",
       "      <td>4.906885</td>\n",
       "      <td>0.395681</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>uniform</td>\n",
       "      <td>{'n_neighbors': 10, 'p': 1, 'weights': 'uniform'}</td>\n",
       "      <td>0.720341</td>\n",
       "      <td>0.730168</td>\n",
       "      <td>...</td>\n",
       "      <td>0.734075</td>\n",
       "      <td>0.727445</td>\n",
       "      <td>0.732063</td>\n",
       "      <td>0.734312</td>\n",
       "      <td>0.729695</td>\n",
       "      <td>0.727208</td>\n",
       "      <td>0.723150</td>\n",
       "      <td>0.728057</td>\n",
       "      <td>0.004675</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 21 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   mean_fit_time  std_fit_time  mean_score_time  std_score_time  \\\n",
       "0       0.231523      0.036805         3.493722        0.971770   \n",
       "1       0.286101      0.051702         4.142222        1.026447   \n",
       "2       0.237938      0.026114         2.596392        0.446166   \n",
       "3       0.235935      0.039246         1.974733        0.192140   \n",
       "4       0.224505      0.032128         4.906885        0.395681   \n",
       "\n",
       "  param_n_neighbors param_p param_weights  \\\n",
       "0                 3       1       uniform   \n",
       "1                 3       1      distance   \n",
       "2                 3       2       uniform   \n",
       "3                 3       2      distance   \n",
       "4                10       1       uniform   \n",
       "\n",
       "                                              params  split0_test_score  \\\n",
       "0   {'n_neighbors': 3, 'p': 1, 'weights': 'uniform'}           0.850343   \n",
       "1  {'n_neighbors': 3, 'p': 1, 'weights': 'distance'}           0.860289   \n",
       "2   {'n_neighbors': 3, 'p': 2, 'weights': 'uniform'}           0.853422   \n",
       "3  {'n_neighbors': 3, 'p': 2, 'weights': 'distance'}           0.862775   \n",
       "4  {'n_neighbors': 10, 'p': 1, 'weights': 'uniform'}           0.720341   \n",
       "\n",
       "   split1_test_score  ...  split3_test_score  split4_test_score  \\\n",
       "0           0.855079  ...           0.849870           0.846199   \n",
       "1           0.864906  ...           0.861236           0.855553   \n",
       "2           0.856855  ...           0.850343           0.850225   \n",
       "3           0.866682  ...           0.861710           0.859578   \n",
       "4           0.730168  ...           0.734075           0.727445   \n",
       "\n",
       "   split5_test_score  split6_test_score  split7_test_score  split8_test_score  \\\n",
       "0           0.853067           0.854724           0.852948           0.857684   \n",
       "1           0.863012           0.868932           0.862420           0.867630   \n",
       "2           0.852119           0.852475           0.850699           0.857447   \n",
       "3           0.861354           0.866682           0.860052           0.867630   \n",
       "4           0.732063           0.734312           0.729695           0.727208   \n",
       "\n",
       "   split9_test_score  mean_test_score  std_test_score  rank_test_score  \n",
       "0           0.850562         0.852402        0.003082                4  \n",
       "1           0.859207         0.862738        0.003764                2  \n",
       "2           0.852694         0.853325        0.002660                3  \n",
       "3           0.861812         0.863673        0.003163                1  \n",
       "4           0.723150         0.728057        0.004675                7  \n",
       "\n",
       "[5 rows x 21 columns]"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "SEED=42\n",
    "np.random.seed(SEED)\n",
    "\n",
    "espaco_de_parametros = {\n",
    "    \"n_neighbors\" : [3, 10],\n",
    "    \"weights\": ['uniform', 'distance'],\n",
    "    \"p\": [1, 2]}\n",
    "\n",
    "busca = GridSearchCV(Knc,\n",
    "                    espaco_de_parametros,\n",
    "                    cv = cv)\n",
    "\n",
    "busca.fit(x_treino, y_treino)\n",
    "resultados = pd.DataFrame(busca.cv_results_)\n",
    "resultados.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "4784904f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'n_neighbors': 3, 'p': 2, 'weights': 'distance'}\n",
      "86.36734763405506\n"
     ]
    }
   ],
   "source": [
    "print(busca.best_params_)\n",
    "print(busca.best_score_ * 100)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ba8754de",
   "metadata": {},
   "source": [
    "### Anotações\n",
    "O GridSerchCV faz o que eu fiz à mão, uma busca pelos parâmentros selecionado realizando todas as permutações de possibilidades das mesmas. E a realização do dataframe para observars os detalhes de cada resultado é um ótima ideia para entender como a variação dos hiperpâmetros influencia a acurácia.\n",
    "\n",
    "**Aprendizado para o futuro:** Utilize o GridSerchCV para quando quiser fazer buscos permutadas entre 2 ou mais hiperparâmetros. Para apenas 1 hiperparâmetro vale a pena fazer da mesma forma que fiz anteriormente.  \n",
    "\n",
    "# Pickle e Deploy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "27c175dc",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "cd581762",
   "metadata": {},
   "outputs": [],
   "source": [
    "file = open('Classifica_musica.pkl', 'wb')\n",
    "pickle.dump(Knc ,file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9cecc4e6",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}