05.03-twitter-classificação-sentimentos-liguagem-natural.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# importação das bibliotecas necessarias\n",
    "\n",
    "from sklearn.metrics import accuracy_score        # acuracia do modelo (avaliação)\n",
    "from sklearn.metrics import confusion_matrix      # matriz confusão (avaliação)\n",
    "from spacy.lang.pt import STOP_WORDS              # stop_word\n",
    "import matplotlib.pyplot as plt                   # visualização   \n",
    "import pandas as pd                               # manipulação dados\n",
    "import numpy as np                                # calculos numericos\n",
    "import seaborn as sns                             # visualização\n",
    "import string                                     # manipulação strings\n",
    "import spacy                                      # liguagem natural\n",
    "import random                                     # valores aleatórios\n",
    "import re                                         # expressão regular\n",
    "\n",
    "# ignorar os warnings de depreciação de metodos\n",
    "import warnings\n",
    "warnings.simplefilter(\"ignore\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# carregamento da base de dados\n",
    "\n",
    "bd_treino = pd.read_csv('./data/twitter_train.csv', delimiter=';')\n",
    "bd_teste = pd.read_csv('./data/twitter_test.csv', delimiter=';')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "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>id</th>\n",
       "      <th>tweet_text</th>\n",
       "      <th>tweet_date</th>\n",
       "      <th>sentiment</th>\n",
       "      <th>query_used</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1050785521201541121</td>\n",
       "      <td>@Laranjito76 A pessoa certa para isso seria o ...</td>\n",
       "      <td>Fri Oct 12 16:29:25 +0000 2018</td>\n",
       "      <td>1</td>\n",
       "      <td>:)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1050785431955140608</td>\n",
       "      <td>@behin_d_curtain Para mim, é precisamente o co...</td>\n",
       "      <td>Fri Oct 12 16:29:04 +0000 2018</td>\n",
       "      <td>1</td>\n",
       "      <td>:)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1050785401248645120</td>\n",
       "      <td>Vou fazer um video hoje... estou pensando em f...</td>\n",
       "      <td>Fri Oct 12 16:28:56 +0000 2018</td>\n",
       "      <td>1</td>\n",
       "      <td>:)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1050785370982547461</td>\n",
       "      <td>aaaaaaaa amei tanto essas polaroids, nem sei e...</td>\n",
       "      <td>Fri Oct 12 16:28:49 +0000 2018</td>\n",
       "      <td>1</td>\n",
       "      <td>:)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1050785368902131713</td>\n",
       "      <td>Valoriza o coração do menininho que vc tem. El...</td>\n",
       "      <td>Fri Oct 12 16:28:49 +0000 2018</td>\n",
       "      <td>1</td>\n",
       "      <td>:)</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                    id                                         tweet_text  \\\n",
       "0  1050785521201541121  @Laranjito76 A pessoa certa para isso seria o ...   \n",
       "1  1050785431955140608  @behin_d_curtain Para mim, é precisamente o co...   \n",
       "2  1050785401248645120  Vou fazer um video hoje... estou pensando em f...   \n",
       "3  1050785370982547461  aaaaaaaa amei tanto essas polaroids, nem sei e...   \n",
       "4  1050785368902131713  Valoriza o coração do menininho que vc tem. El...   \n",
       "\n",
       "                       tweet_date  sentiment query_used  \n",
       "0  Fri Oct 12 16:29:25 +0000 2018          1         :)  \n",
       "1  Fri Oct 12 16:29:04 +0000 2018          1         :)  \n",
       "2  Fri Oct 12 16:28:56 +0000 2018          1         :)  \n",
       "3  Fri Oct 12 16:28:49 +0000 2018          1         :)  \n",
       "4  Fri Oct 12 16:28:49 +0000 2018          1         :)  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# primeiros registros\n",
    "\n",
    "bd_treino.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "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>id</th>\n",
       "      <th>tweet_text</th>\n",
       "      <th>tweet_date</th>\n",
       "      <th>sentiment</th>\n",
       "      <th>query_used</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1029536486021099522</td>\n",
       "      <td>@Gazo1a Nossa! Muito obrigada :)</td>\n",
       "      <td>Wed Aug 15 01:13:20 +0000 2018</td>\n",
       "      <td>1</td>\n",
       "      <td>:)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1029536496368406528</td>\n",
       "      <td>@BerzGamer vai pa puta que te pariu :)</td>\n",
       "      <td>Wed Aug 15 01:13:23 +0000 2018</td>\n",
       "      <td>1</td>\n",
       "      <td>:)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1029536531655131137</td>\n",
       "      <td>QUER MAIS DESCONTOS? (14/08) ⭐⭐⭐⭐⭐ 🌐 Confira n...</td>\n",
       "      <td>Wed Aug 15 01:13:31 +0000 2018</td>\n",
       "      <td>1</td>\n",
       "      <td>:)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1029536560117678081</td>\n",
       "      <td>EU VOU PEGAR VCS, ME AJUDEM GALERA, PELO AMOR ...</td>\n",
       "      <td>Wed Aug 15 01:13:38 +0000 2018</td>\n",
       "      <td>1</td>\n",
       "      <td>:)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1029536605852377088</td>\n",
       "      <td>Estávamos em casa do Zé e eu estava a morrer d...</td>\n",
       "      <td>Wed Aug 15 01:13:49 +0000 2018</td>\n",
       "      <td>1</td>\n",
       "      <td>:)</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                    id                                         tweet_text  \\\n",
       "0  1029536486021099522                   @Gazo1a Nossa! Muito obrigada :)   \n",
       "1  1029536496368406528             @BerzGamer vai pa puta que te pariu :)   \n",
       "2  1029536531655131137  QUER MAIS DESCONTOS? (14/08) ⭐⭐⭐⭐⭐ 🌐 Confira n...   \n",
       "3  1029536560117678081  EU VOU PEGAR VCS, ME AJUDEM GALERA, PELO AMOR ...   \n",
       "4  1029536605852377088  Estávamos em casa do Zé e eu estava a morrer d...   \n",
       "\n",
       "                       tweet_date  sentiment query_used  \n",
       "0  Wed Aug 15 01:13:20 +0000 2018          1         :)  \n",
       "1  Wed Aug 15 01:13:23 +0000 2018          1         :)  \n",
       "2  Wed Aug 15 01:13:31 +0000 2018          1         :)  \n",
       "3  Wed Aug 15 01:13:38 +0000 2018          1         :)  \n",
       "4  Wed Aug 15 01:13:49 +0000 2018          1         :)  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# primeiros registros teste\n",
    "\n",
    "bd_teste.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# visualização dos grupos ['emocao']\n",
    "\n",
    "sns.countplot(bd_treino['sentiment'], label='Contagem');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# pre processamento de texto"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# retirando os atributos que não usados\n",
    "\n",
    "bd_treino.drop(['id', 'tweet_date', 'query_used'], inplace=True, axis=1)\n",
    "bd_teste.drop(['id', 'tweet_date', 'query_used'], inplace=True, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "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>tweet_text</th>\n",
       "      <th>sentiment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>@Laranjito76 A pessoa certa para isso seria o ...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>@behin_d_curtain Para mim, é precisamente o co...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Vou fazer um video hoje... estou pensando em f...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>aaaaaaaa amei tanto essas polaroids, nem sei e...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Valoriza o coração do menininho que vc tem. El...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                          tweet_text  sentiment\n",
       "0  @Laranjito76 A pessoa certa para isso seria o ...          1\n",
       "1  @behin_d_curtain Para mim, é precisamente o co...          1\n",
       "2  Vou fazer um video hoje... estou pensando em f...          1\n",
       "3  aaaaaaaa amei tanto essas polaroids, nem sei e...          1\n",
       "4  Valoriza o coração do menininho que vc tem. El...          1"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# viz após treinamento\n",
    "\n",
    "bd_treino.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "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>tweet_text</th>\n",
       "      <th>sentiment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>@Gazo1a Nossa! Muito obrigada :)</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>@BerzGamer vai pa puta que te pariu :)</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>QUER MAIS DESCONTOS? (14/08) ⭐⭐⭐⭐⭐ 🌐 Confira n...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>EU VOU PEGAR VCS, ME AJUDEM GALERA, PELO AMOR ...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Estávamos em casa do Zé e eu estava a morrer d...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                          tweet_text  sentiment\n",
       "0                   @Gazo1a Nossa! Muito obrigada :)          1\n",
       "1             @BerzGamer vai pa puta que te pariu :)          1\n",
       "2  QUER MAIS DESCONTOS? (14/08) ⭐⭐⭐⭐⭐ 🌐 Confira n...          1\n",
       "3  EU VOU PEGAR VCS, ME AJUDEM GALERA, PELO AMOR ...          1\n",
       "4  Estávamos em casa do Zé e eu estava a morrer d...          1"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bd_teste.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# viz grafico\n",
    "\n",
    "sns.heatmap(pd.isnull(bd_treino));"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.heatmap(pd.isnull(bd_teste));"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<spacy.lang.pt.Portuguese at 0x24126e45580>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# modelo em portugues\n",
    "\n",
    "pln = spacy.load('pt')\n",
    "pln"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# stop_words\n",
    "\n",
    "stop_words = STOP_WORDS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "def preprocessamento(text):\n",
    "  # Letras minúsculas\n",
    "  text = text.lower()\n",
    "\n",
    "  # Nome do usuário\n",
    "  text = re.sub(r\"@[A-Za-z0-9$-_@.&+]+\", ' ', text)\n",
    "\n",
    "  # URLs\n",
    "  text = re.sub(r\"https?://[A-Za-z0-9./]+\", ' ', text)\n",
    "\n",
    "  # Espaços em branco\n",
    "  text = re.sub(r\" +\", ' ', text)\n",
    "\n",
    "  # Emoticons\n",
    "  lista_emocoes = {':)': 'emocaopositiva',\n",
    "                   ':d': 'emocaopositiva',\n",
    "                   ':(': 'emocaonegativa'}\n",
    "  for emocao in lista_emocoes:\n",
    "    text = text.replace(emocao, lista_emocoes[emocao])\n",
    "\n",
    "  # Lematização\n",
    "  documento = pln(text)\n",
    "\n",
    "  lista = []\n",
    "  for token in documento:\n",
    "    lista.append(token.lemma_)\n",
    "  \n",
    "  # Stop words e pontuações\n",
    "  lista = [palavra for palavra in lista if palavra not in stop_words and palavra not in string.punctuation]\n",
    "  lista = ' '.join([str(elemento) for elemento in lista if not elemento.isdigit()])\n",
    "  \n",
    "  return lista"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# aplicando preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "# tratamento com a função preprocessamento\n",
    "\n",
    "bd_treino['tweet_text'] = bd_treino['tweet_text'].apply(preprocessamento)\n",
    "bd_teste['tweet_text'] = bd_teste['tweet_text'].apply(preprocessamento)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'aprender liguagem natural python brasilia'"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# viz após preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "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>tweet_text</th>\n",
       "      <th>sentiment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>o pessoa certo parir seriar o valer e azeved...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>parir mim precisamente o contrário emocaopos...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>video hoje ... pensar falar sobrar o meter csg...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>aaaaaaaa amar polaroids expressar o quantum to...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>valorizar o coração menino vc diferente o sorr...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                          tweet_text  sentiment\n",
       "0    o pessoa certo parir seriar o valer e azeved...          1\n",
       "1    parir mim precisamente o contrário emocaopos...          1\n",
       "2  video hoje ... pensar falar sobrar o meter csg...          1\n",
       "3  aaaaaaaa amar polaroids expressar o quantum to...          1\n",
       "4  valorizar o coração menino vc diferente o sorr...          1"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bd_treino.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "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>tweet_text</th>\n",
       "      <th>sentiment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>obrigar emocaopositiva</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>pa puta parir emocaopositiva</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>querer desconto 14/08 ⭐ ⭐ ⭐ ⭐ ⭐ 🌐 conferir o l...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>vcs ajudar galera pelar amor butera emocaoposi...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>casar zé e o morrer sono chegar o casar e fica...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                          tweet_text  sentiment\n",
       "0                             obrigar emocaopositiva          1\n",
       "1                       pa puta parir emocaopositiva          1\n",
       "2  querer desconto 14/08 ⭐ ⭐ ⭐ ⭐ ⭐ 🌐 conferir o l...          1\n",
       "3  vcs ajudar galera pelar amor butera emocaoposi...          1\n",
       "4  casar zé e o morrer sono chegar o casar e fica...          1"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bd_teste.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "base_dados_treinamento_final = []\n",
    "\n",
    "for texto, emocao in zip(bd_treino['tweet_text'], bd_treino['sentiment']):\n",
    "  \n",
    "  if emocao == 1:\n",
    "    dic = ({'POSITIVO': True, 'NEGATIVO': False})\n",
    "  \n",
    "  elif emocao == 0:\n",
    "    dic = ({'POSITIVO': False, 'NEGATIVO': True})\n",
    "\n",
    "  base_dados_treinamento_final.append([texto, dic.copy()])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[['  amigo to emocaopositiva', {'POSITIVO': True, 'NEGATIVO': False}],\n",
       " ['  bravo jairo emocaopositiva', {'POSITIVO': True, 'NEGATIVO': False}],\n",
       " ['  socorrer ksoaksosks achar to o caixa som :p',\n",
       "  {'POSITIVO': True, 'NEGATIVO': False}],\n",
       " ['  gostar fazenda emocaopositiva morrer medo galinha kkkkkkk',\n",
       "  {'POSITIVO': True, 'NEGATIVO': False}],\n",
       " ['amanhar venom emocaopositiva', {'POSITIVO': True, 'NEGATIVO': False}]]"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# viz após tratameto\n",
    "\n",
    "base_dados_treinamento_final[5:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "# criação classificador"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "# parametros do modelo\n",
    "\n",
    "model = spacy.blank('pt')\n",
    "\n",
    "categorias = model.create_pipe('textcat')\n",
    "categorias.add_label('POSITIVO')\n",
    "categorias.add_label('NEGATIVO')\n",
    "model.add_pipe(categorias)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'textcat': 6.629547314986505e-07}\n"
     ]
    }
   ],
   "source": [
    "# treinamento do modelo\n",
    "\n",
    "historico = []\n",
    "model.begin_training()\n",
    "\n",
    "for epoch in range(5):\n",
    "    \n",
    "    random.shuffle(base_dados_treinamento_final)\n",
    "    losses = {}\n",
    "\n",
    "    for batch in spacy.util.minibatch(base_dados_treinamento_final, 512):\n",
    "        \n",
    "        textos = [model(texto) for texto, entities in batch]\n",
    "        annotations = [{'cats':entities} for texto, entities in batch]\n",
    "        model.update(textos, annotations, losses=losses)\n",
    "        historico.append(losses)\n",
    "\n",
    "    if epoch % 5 == 0:\n",
    "        print(losses)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    " # lista com losses do modelo\n",
    "\n",
    "historico_loss = []\n",
    "for i in historico:\n",
    "    historico_loss.append(i.get('textcat'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 6.62954731e-07, 6.62954731e-07,\n",
       "       6.62954731e-07, 6.62954731e-07, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       1.49689923e-06, 1.49689923e-06, 1.49689923e-06, 1.49689923e-06,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 8.83931614e-07, 8.83931614e-07,\n",
       "       8.83931614e-07, 8.83931614e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.69718628e-07, 4.69718628e-07, 4.69718628e-07, 4.69718628e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07, 4.44035603e-07, 4.44035603e-07,\n",
       "       4.44035603e-07, 4.44035603e-07])"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# conversão para array a lista de erros\n",
    "\n",
    "historico_loss = np.array(historico_loss)\n",
    "historico_loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# viz erro\n",
    "\n",
    "plt.plot(historico_loss)\n",
    "plt.title('Progressão do erro')\n",
    "plt.xlabel('Batches')\n",
    "plt.ylabel('Erro')\n",
    "plt.show();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "# salvando modelo\n",
    "\n",
    "model.to_disk('./data/modelo_twitter')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<spacy.lang.pt.Portuguese at 0x24162412520>"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# carregando modelo\n",
    "\n",
    "model_carregado = spacy.load('./data/modelo_twitter')\n",
    "model_carregado"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'  ficar feliz q recíproco emocaopositiva'"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# texto positivo\n",
    "\n",
    "texto_positivo = bd_teste['tweet_text'][21]\n",
    "texto_positivo"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "  ficar feliz q recíproco emocaopositiva"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# previsão\n",
    "\n",
    "prev = model_carregado(texto_positivo)\n",
    "prev"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'POSITIVO': 0.9999545812606812, 'NEGATIVO': 4.539787187241018e-05}"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# categorias da previsão\n",
    "\n",
    "prev.cats"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'gostar'"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "texto_positivo = 'eu gosto muito de você'\n",
    "texto_positivo = preprocessamento(texto_positivo)\n",
    "texto_positivo"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'POSITIVO': 0.012623820453882217, 'NEGATIVO': 0.9943988919258118}"
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_carregado(texto_positivo).cats"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'tô triste filmar mulan o mushu emocaonegativa'"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# texto negativo\n",
    "\n",
    "bd_teste['tweet_text'][4000]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'POSITIVO': 5.4001608077669516e-05, 'NEGATIVO': 0.9999545812606812}"
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# previsão texto negativo\n",
    "\n",
    "texto_negativo = bd_teste['tweet_text'][4000]\n",
    "previsao = model_carregado(texto_negativo)\n",
    "previsao.cats"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Avaliação do modelo"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [],
   "source": [
    "# previsões base teste\n",
    "\n",
    "previsoes = []\n",
    "\n",
    "for texto in bd_teste['tweet_text']:\n",
    "\n",
    "  previsao = model_carregado(texto)\n",
    "  previsoes.append(previsao.cats)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "# previsao base_teste\n",
    "\n",
    "previsoes_final = []\n",
    "\n",
    "for previsao in previsoes:\n",
    "\n",
    "  if previsao['POSITIVO'] > previsao['NEGATIVO']:\n",
    "    previsoes_final.append(1)\n",
    "  else:\n",
    "    previsoes_final.append(0)\n",
    "\n",
    "previsoes_final = np.array(previsoes_final)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "# conversão array\n",
    "\n",
    "respostas_reais = bd_teste['sentiment'].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9976"
      ]
     },
     "execution_count": 73,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# acuracia modelo\n",
    "\n",
    "accuracy_score(respostas_reais, previsoes_final)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# matriz de confusão\n",
    "\n",
    "cm = confusion_matrix(respostas_reais, previsoes_final)\n",
    "sns.heatmap(cm, annot=True);"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "ad2bdc8ecc057115af97d19610ffacc2b4e99fae6737bb82f5d7fb13d2f2c186"
  },
  "kernelspec": {
   "display_name": "Python 3.8.8 ('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": "3.8.8"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}