Machine Learning - Dados heterogêneos, Text Mining Raiz.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Minerando Dados\n",
    "* Rodrigo Santana \n",
    "* E-mail para contato: contato@minerandodados.com.br\n",
    "* Site: www.minerandodados.com.br\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 2,
     "metadata": {
      "image/png": {
       "height": 200,
       "width": 300
      }
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from IPython.display import Image\n",
    "Image('minerador.png',width=300, height=200)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "** Importando as bibliotecas necessárias**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "from sklearn.naive_bayes import MultinomialNB\n",
    "from sklearn import metrics\n",
    "from sklearn.model_selection import cross_val_predict"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Ler arquivo de dados e conta a quantidade de linhas**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = pd.read_csv('titulo_noticias.txt',encoding='utf-8')"
   ]
  },
  {
   "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>noticia</th>\n",
       "      <th>sentimento</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>diretor petrobras nega organização criminosa e...</td>\n",
       "      <td>positivo</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>tom cauteloso janet yellen pressiona bolsas f...</td>\n",
       "      <td>neutro</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>bovespa caminha nova máxima ano quarta alta s...</td>\n",
       "      <td>positivo</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>após abrir estável ibovespa passa registrar q...</td>\n",
       "      <td>negativo</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>paulo miranda reivindica iluminação telefonia...</td>\n",
       "      <td>positivo</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                             noticia sentimento\n",
       "0  diretor petrobras nega organização criminosa e...   positivo\n",
       "1   tom cauteloso janet yellen pressiona bolsas f...     neutro\n",
       "2   bovespa caminha nova máxima ano quarta alta s...   positivo\n",
       "3   após abrir estável ibovespa passa registrar q...   negativo\n",
       "4   paulo miranda reivindica iluminação telefonia...   positivo"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset.head(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "noticia       2123\n",
       "sentimento    2123\n",
       "dtype: int64"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset.count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x23ceeedce48>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x23ceed2c748>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "dataset.sentimento.value_counts().plot(kind='bar')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Distribuição das classes neutros, positivos e negativos**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "** Modelo Simples**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.pipeline import Pipeline\n",
    "\n",
    "pip_simples = Pipeline([\n",
    "  ('counts', CountVectorizer()),\n",
    "  ('classifier', MultinomialNB())\n",
    "])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Função de avaliação de modelos**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def avalia_modelo(clf, X, y):\n",
    "    resultados = cross_val_predict(clf, X, y, cv=5)\n",
    "    print (pd.crosstab(y, resultados, rownames=['Real'], colnames=['Predito'], margins=True))\n",
    "    return np.mean(cross_val_score(clf, X, y, cv=5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predito     negativo   neutro   positivo   All\n",
      "Real                                          \n",
      " negativo        531       39        176   746\n",
      " neutro          103      142        207   452\n",
      " positivo        111       63        751   925\n",
      "All              745      244       1134  2123\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.6707639451209663"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "avalia_modelo(pip_simples,dataset.noticia,dataset.sentimento)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "** Função para tag de Negações**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def marque_negacao(texto):\n",
    "    negacoes = ['não','not']\n",
    "    negacao_detectada = False\n",
    "    resultado = []\n",
    "    palavras = texto.split()\n",
    "    for p in palavras:\n",
    "        p = p.lower()\n",
    "        if negacao_detectada == True:\n",
    "            p = p + '_NEG'\n",
    "        if p in negacoes:\n",
    "            negacao_detectada = True\n",
    "        resultado.append(p)\n",
    "    return resultado"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "** Visualizando a função de negações**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['eu', 'gosto', 'do', 'brasil']\n",
      "['eu', 'não', 'gosto_NEG', 'do_NEG', 'brasil_NEG']\n"
     ]
    }
   ],
   "source": [
    "print (marque_negacao('Eu gosto do Brasil'))\n",
    "print (marque_negacao('Eu não gosto do Brasil'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "** Cria o pipeline que utiliza a função de negações**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "pip_neg = Pipeline([\n",
    "  ('counts', CountVectorizer(tokenizer=lambda text: marque_negacao(text))),\n",
    "  ('classifier', MultinomialNB())\n",
    "])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Avaliando o modelo com negações**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predito     negativo   neutro   positivo   All\n",
      "Real                                          \n",
      " negativo        520       43        183   746\n",
      " neutro          105      142        205   452\n",
      " positivo        115       64        746   925\n",
      "All              740      249       1134  2123\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.6632311724583536"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "avalia_modelo(pip_neg,dataset.noticia,dataset.sentimento)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# **União de Features**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* Classes para fazer união de features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.base import BaseEstimator, TransformerMixin\n",
    "\n",
    "class QtdPalavras(BaseEstimator, TransformerMixin):\n",
    "    \n",
    "    def __init__(self):\n",
    "        pass\n",
    "    \n",
    "    def tamanho_noticia(self, texto):\n",
    "        return len(texto.split())\n",
    "    \n",
    "    def transform(self, news, y=None):\n",
    "        vet_tam = []\n",
    "        vet_tam_2 = []\n",
    "        for n in news:\n",
    "            vet_tam.append(self.tamanho_noticia(n))\n",
    "        vet_tam_2 = [[i] for i in vet_tam]\n",
    "        return vet_tam_2\n",
    "    \n",
    "    def fit(self, df, y=None):\n",
    "        return self"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.pipeline import FeatureUnion"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "features = FeatureUnion([('QuantidadePalavras',QtdPalavras())])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "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>noticia</th>\n",
       "      <th>sentimento</th>\n",
       "      <th>qtd_palavras</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>diretor petrobras nega organização criminosa e...</td>\n",
       "      <td>positivo</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>tom cauteloso janet yellen pressiona bolsas f...</td>\n",
       "      <td>neutro</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>bovespa caminha nova máxima ano quarta alta s...</td>\n",
       "      <td>positivo</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>após abrir estável ibovespa passa registrar q...</td>\n",
       "      <td>negativo</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>paulo miranda reivindica iluminação telefonia...</td>\n",
       "      <td>positivo</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                             noticia sentimento  qtd_palavras\n",
       "0  diretor petrobras nega organização criminosa e...   positivo             8\n",
       "1   tom cauteloso janet yellen pressiona bolsas f...     neutro             8\n",
       "2   bovespa caminha nova máxima ano quarta alta s...   positivo             8\n",
       "3   após abrir estável ibovespa passa registrar q...   negativo             9\n",
       "4   paulo miranda reivindica iluminação telefonia...   positivo             7"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset['qtd_palavras'] = features.fit_transform(dataset.noticia,dataset.sentimento)\n",
    "dataset.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset.drop(['qtd_palavras'], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Cria classes para uniao de features**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "** Cria o primeiro pipeline para processamento do texto**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<2123x3803 sparse matrix of type '<class 'numpy.int64'>'\n",
       "\twith 17325 stored elements in Compressed Sparse Row format>"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "text_process = Pipeline([\n",
    "                ('count', CountVectorizer())\n",
    "            ])\n",
    "text_process.fit_transform(dataset.noticia)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "** Cria o segundo pipeline para a segunda feature**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[[8],\n",
       " [8],\n",
       " [8],\n",
       " [9],\n",
       " [7],\n",
       " [8],\n",
       " [10],\n",
       " [7],\n",
       " [10],\n",
       " [6],\n",
       " [9],\n",
       " [14],\n",
       " [11],\n",
       " [8],\n",
       " [15],\n",
       " [7],\n",
       " [6],\n",
       " [10],\n",
       " [11],\n",
       " [6],\n",
       " [9],\n",
       " [11],\n",
       " [7],\n",
       " [7],\n",
       " [10],\n",
       " [7],\n",
       " [9],\n",
       " [10],\n",
       " [10],\n",
       " [5],\n",
       " [5],\n",
       " [8],\n",
       " [7],\n",
       " [7],\n",
       " [10],\n",
       " [7],\n",
       " [8],\n",
       " [10],\n",
       " [6],\n",
       " [11],\n",
       " [10],\n",
       " [8],\n",
       " [6],\n",
       " [7],\n",
       " [10],\n",
       " [9],\n",
       " [8],\n",
       " [9],\n",
       " [7],\n",
       " [6],\n",
       " [8],\n",
       " [8],\n",
       " [9],\n",
       " [9],\n",
       " [4],\n",
       " [14],\n",
       " [9],\n",
       " [13],\n",
       " [8],\n",
       " [9],\n",
       " [11],\n",
       " [6],\n",
       " [6],\n",
       " [4],\n",
       " [9],\n",
       " [4],\n",
       " [10],\n",
       " [7],\n",
       " [2],\n",
       " [7],\n",
       " [12],\n",
       " [6],\n",
       " [9],\n",
       " [9],\n",
       " [10],\n",
       " [8],\n",
       " [6],\n",
       " [7],\n",
       " [12],\n",
       " [10],\n",
       " [6],\n",
       " [11],\n",
       " [4],\n",
       " [13],\n",
       " [8],\n",
       " [7],\n",
       " [6],\n",
       " [9],\n",
       " [8],\n",
       " [7],\n",
       " [10],\n",
       " [12],\n",
       " [3],\n",
       " [10],\n",
       " [7],\n",
       " [8],\n",
       " [7],\n",
       " [6],\n",
       " [7],\n",
       " [11],\n",
       " [10],\n",
       " [7],\n",
       " [5],\n",
       " [5],\n",
       " [8],\n",
       " [7],\n",
       " [9],\n",
       " [7],\n",
       " [7],\n",
       " [5],\n",
       " [10],\n",
       " [15],\n",
       " [8],\n",
       " [8],\n",
       " [11],\n",
       " [6],\n",
       " [6],\n",
       " [2],\n",
       " [7],\n",
       " [5],\n",
       " [9],\n",
       " [7],\n",
       " [2],\n",
       " [7],\n",
       " [5],\n",
       " [7],\n",
       " [8],\n",
       " [7],\n",
       " [11],\n",
       " [8],\n",
       " [4],\n",
       " [8],\n",
       " [8],\n",
       " [6],\n",
       " [11],\n",
       " [9],\n",
       " [7],\n",
       " [6],\n",
       " [9],\n",
       " [6],\n",
       " [7],\n",
       " [13],\n",
       " [10],\n",
       " [4],\n",
       " [10],\n",
       " [9],\n",
       " [12],\n",
       " [7],\n",
       " [7],\n",
       " [9],\n",
       " [8],\n",
       " [12],\n",
       " [10],\n",
       " [8],\n",
       " [11],\n",
       " [9],\n",
       " [9],\n",
       " [9],\n",
       " [5],\n",
       " [9],\n",
       " [10],\n",
       " [8],\n",
       " [8],\n",
       " [9],\n",
       " [12],\n",
       " [6],\n",
       " [8],\n",
       " [8],\n",
       " [6],\n",
       " [13],\n",
       " [8],\n",
       " [7],\n",
       " [7],\n",
       " [11],\n",
       " [5],\n",
       " [6],\n",
       " [5],\n",
       " [8],\n",
       " [5],\n",
       " [4],\n",
       " [5],\n",
       " [9],\n",
       " [8],\n",
       " [7],\n",
       " [11],\n",
       " [6],\n",
       " [2],\n",
       " [6],\n",
       " [10],\n",
       " [7],\n",
       " [12],\n",
       " [12],\n",
       " [12],\n",
       " [5],\n",
       " [9],\n",
       " [7],\n",
       " [5],\n",
       " [11],\n",
       " [10],\n",
       " [12],\n",
       " [9],\n",
       " [6],\n",
       " [7],\n",
       " [9],\n",
       " [11],\n",
       " [9],\n",
       " [10],\n",
       " [7],\n",
       " [8],\n",
       " [10],\n",
       " [8],\n",
       " [8],\n",
       " [10],\n",
       " [6],\n",
       " [7],\n",
       " [7],\n",
       " [11],\n",
       " [9],\n",
       " [12],\n",
       " [9],\n",
       " [6],\n",
       " [7],\n",
       " [7],\n",
       " [7],\n",
       " [8],\n",
       " [5],\n",
       " [10],\n",
       " [9],\n",
       " [7],\n",
       " [9],\n",
       " [9],\n",
       " [6],\n",
       " [8],\n",
       " [11],\n",
       " [8],\n",
       " [6],\n",
       " [5],\n",
       " [7],\n",
       " [5],\n",
       " [7],\n",
       " [7],\n",
       " [7],\n",
       " [7],\n",
       " [5],\n",
       " [11],\n",
       " [10],\n",
       " [12],\n",
       " [6],\n",
       " [11],\n",
       " [8],\n",
       " [6],\n",
       " [10],\n",
       " [8],\n",
       " [6],\n",
       " [7],\n",
       " [6],\n",
       " [10],\n",
       " [5],\n",
       " [7],\n",
       " [5],\n",
       " [6],\n",
       " [9],\n",
       " [9],\n",
       " [10],\n",
       " [7],\n",
       " [6],\n",
       " [9],\n",
       " [10],\n",
       " [9],\n",
       " [9],\n",
       " [10],\n",
       " [6],\n",
       " [12],\n",
       " [7],\n",
       " [9],\n",
       " [7],\n",
       " [9],\n",
       " [9],\n",
       " [8],\n",
       " [9],\n",
       " [6],\n",
       " [7],\n",
       " [3],\n",
       " [8],\n",
       " [6],\n",
       " [8],\n",
       " [10],\n",
       " [6],\n",
       " [7],\n",
       " [4],\n",
       " [6],\n",
       " [8],\n",
       " [9],\n",
       " [12],\n",
       " [9],\n",
       " [10],\n",
       " [5],\n",
       " [11],\n",
       " [8],\n",
       " [9],\n",
       " [7],\n",
       " [6],\n",
       " [9],\n",
       " [8],\n",
       " [11],\n",
       " [8],\n",
       " [11],\n",
       " [9],\n",
       " [7],\n",
       " [10],\n",
       " [8],\n",
       " [5],\n",
       " [9],\n",
       " [9],\n",
       " [11],\n",
       " [10],\n",
       " [8],\n",
       " [5],\n",
       " [9],\n",
       " [8],\n",
       " [12],\n",
       " [10],\n",
       " [11],\n",
       " [10],\n",
       " [9],\n",
       " [10],\n",
       " [8],\n",
       " [8],\n",
       " [5],\n",
       " [10],\n",
       " [9],\n",
       " [10],\n",
       " [7],\n",
       " [9],\n",
       " [6],\n",
       " [9],\n",
       " [10],\n",
       " [5],\n",
       " [7],\n",
       " [5],\n",
       " [9],\n",
       " [6],\n",
       " [9],\n",
       " [10],\n",
       " [10],\n",
       " [5],\n",
       " [5],\n",
       " [5],\n",
       " [4],\n",
       " [7],\n",
       " [10],\n",
       " [7],\n",
       " [8],\n",
       " [4],\n",
       " [6],\n",
       " [8],\n",
       " [8],\n",
       " [10],\n",
       " [8],\n",
       " [6],\n",
       " [9],\n",
       " [12],\n",
       " [11],\n",
       " [10],\n",
       " [8],\n",
       " [8],\n",
       " [7],\n",
       " [5],\n",
       " [6],\n",
       " [10],\n",
       " [5],\n",
       " [8],\n",
       " [8],\n",
       " [7],\n",
       " [9],\n",
       " [5],\n",
       " [10],\n",
       " [8],\n",
       " [9],\n",
       " [8],\n",
       " [6],\n",
       " [9],\n",
       " [11],\n",
       " [5],\n",
       " [10],\n",
       " [9],\n",
       " [8],\n",
       " [9],\n",
       " [10],\n",
       " [9],\n",
       " [11],\n",
       " [8],\n",
       " [10],\n",
       " [5],\n",
       " [9],\n",
       " [7],\n",
       " [7],\n",
       " [7],\n",
       " [7],\n",
       " [8],\n",
       " [8],\n",
       " [11],\n",
       " [7],\n",
       " [5],\n",
       " [8],\n",
       " [12],\n",
       " [9],\n",
       " [12],\n",
       " [12],\n",
       " [10],\n",
       " [6],\n",
       " [8],\n",
       " [6],\n",
       " [9],\n",
       " [7],\n",
       " [18],\n",
       " [10],\n",
       " [10],\n",
       " [10],\n",
       " [5],\n",
       " [7],\n",
       " [14],\n",
       " [14],\n",
       " [10],\n",
       " [5],\n",
       " [11],\n",
       " [8],\n",
       " [9],\n",
       " [9],\n",
       " [13],\n",
       " [5],\n",
       " [11],\n",
       " [9],\n",
       " [6],\n",
       " [8],\n",
       " [7],\n",
       " [11],\n",
       " [10],\n",
       " [4],\n",
       " [5],\n",
       " [5],\n",
       " [10],\n",
       " [9],\n",
       " [8],\n",
       " [6],\n",
       " [10],\n",
       " [7],\n",
       " [7],\n",
       " [6],\n",
       " [10],\n",
       " [7],\n",
       " [8],\n",
       " [7],\n",
       " [8],\n",
       " [5],\n",
       " [5],\n",
       " [9],\n",
       " [7],\n",
       " [4],\n",
       " [5],\n",
       " [4],\n",
       " [5],\n",
       " [6],\n",
       " [9],\n",
       " [5],\n",
       " [10],\n",
       " [9],\n",
       " [9],\n",
       " [9],\n",
       " [8],\n",
       " [11],\n",
       " [10],\n",
       " [9],\n",
       " [9],\n",
       " [6],\n",
       " [9],\n",
       " [5],\n",
       " [7],\n",
       " [5],\n",
       " [10],\n",
       " [6],\n",
       " [6],\n",
       " [10],\n",
       " [12],\n",
       " [11],\n",
       " [6],\n",
       " [9],\n",
       " [7],\n",
       " [6],\n",
       " [10],\n",
       " [7],\n",
       " [13],\n",
       " [7],\n",
       " [9],\n",
       " [9],\n",
       " [6],\n",
       " [6],\n",
       " [11],\n",
       " [7],\n",
       " [8],\n",
       " [10],\n",
       " [11],\n",
       " [9],\n",
       " [7],\n",
       " [10],\n",
       " [12],\n",
       " [8],\n",
       " [6],\n",
       " [6],\n",
       " [8],\n",
       " [9],\n",
       " [5],\n",
       " [7],\n",
       " [10],\n",
       " [9],\n",
       " [11],\n",
       " [7],\n",
       " [8],\n",
       " [8],\n",
       " [10],\n",
       " [4],\n",
       " [6],\n",
       " [7],\n",
       " [9],\n",
       " [6],\n",
       " [10],\n",
       " [3],\n",
       " [8],\n",
       " [6],\n",
       " [8],\n",
       " [11],\n",
       " [9],\n",
       " [7],\n",
       " [8],\n",
       " [7],\n",
       " [13],\n",
       " [5],\n",
       " [6],\n",
       " [9],\n",
       " [9],\n",
       " [8],\n",
       " [4],\n",
       " [12],\n",
       " [5],\n",
       " [5],\n",
       " [5],\n",
       " [10],\n",
       " [2],\n",
       " [5],\n",
       " [8],\n",
       " [8],\n",
       " [12],\n",
       " [8],\n",
       " [6],\n",
       " [7],\n",
       " [4],\n",
       " [7],\n",
       " [6],\n",
       " [12],\n",
       " [9],\n",
       " [9],\n",
       " [11],\n",
       " [9],\n",
       " [10],\n",
       " [8],\n",
       " [8],\n",
       " [15],\n",
       " [6],\n",
       " [8],\n",
       " [7],\n",
       " [11],\n",
       " [8],\n",
       " [6],\n",
       " [9],\n",
       " [5],\n",
       " [7],\n",
       " [7],\n",
       " [7],\n",
       " [8],\n",
       " [10],\n",
       " [6],\n",
       " [12],\n",
       " [4],\n",
       " [5],\n",
       " [5],\n",
       " [8],\n",
       " [7],\n",
       " [5],\n",
       " [11],\n",
       " [9],\n",
       " [8],\n",
       " [10],\n",
       " [10],\n",
       " [4],\n",
       " [11],\n",
       " [7],\n",
       " [12],\n",
       " [10],\n",
       " [9],\n",
       " [6],\n",
       " [8],\n",
       " [8],\n",
       " [8],\n",
       " [6],\n",
       " [11],\n",
       " [20],\n",
       " [5],\n",
       " [10],\n",
       " [10],\n",
       " [7],\n",
       " [8],\n",
       " [12],\n",
       " [10],\n",
       " [12],\n",
       " [12],\n",
       " [4],\n",
       " [11],\n",
       " [8],\n",
       " [5],\n",
       " [8],\n",
       " [9],\n",
       " [10],\n",
       " [9],\n",
       " [6],\n",
       " [4],\n",
       " [5],\n",
       " [1],\n",
       " [10],\n",
       " [5],\n",
       " [11],\n",
       " [10],\n",
       " [7],\n",
       " [9],\n",
       " [6],\n",
       " [7],\n",
       " [12],\n",
       " [5],\n",
       " [9],\n",
       " [7],\n",
       " [8],\n",
       " [8],\n",
       " [13],\n",
       " [8],\n",
       " [9],\n",
       " [8],\n",
       " [10],\n",
       " [7],\n",
       " [8],\n",
       " [14],\n",
       " [11],\n",
       " [7],\n",
       " [11],\n",
       " [10],\n",
       " [9],\n",
       " [6],\n",
       " [9],\n",
       " [10],\n",
       " [6],\n",
       " [13],\n",
       " [8],\n",
       " [7],\n",
       " [8],\n",
       " [7],\n",
       " [6],\n",
       " [8],\n",
       " [7],\n",
       " [6],\n",
       " [7],\n",
       " [6],\n",
       " [12],\n",
       " [7],\n",
       " [9],\n",
       " [2],\n",
       " [7],\n",
       " [9],\n",
       " [8],\n",
       " [12],\n",
       " [11],\n",
       " [3],\n",
       " [10],\n",
       " [14],\n",
       " [11],\n",
       " [5],\n",
       " [11],\n",
       " [5],\n",
       " [7],\n",
       " [10],\n",
       " [9],\n",
       " [7],\n",
       " [11],\n",
       " [11],\n",
       " [7],\n",
       " [9],\n",
       " [8],\n",
       " [7],\n",
       " [9],\n",
       " [8],\n",
       " [13],\n",
       " [5],\n",
       " [10],\n",
       " [8],\n",
       " [11],\n",
       " [10],\n",
       " [8],\n",
       " [2],\n",
       " [11],\n",
       " [8],\n",
       " [9],\n",
       " [8],\n",
       " [8],\n",
       " [7],\n",
       " [13],\n",
       " [6],\n",
       " [11],\n",
       " [8],\n",
       " [13],\n",
       " [8],\n",
       " [9],\n",
       " [10],\n",
       " [8],\n",
       " [9],\n",
       " [8],\n",
       " [7],\n",
       " [7],\n",
       " [6],\n",
       " [5],\n",
       " [5],\n",
       " [6],\n",
       " [8],\n",
       " [7],\n",
       " [8],\n",
       " [3],\n",
       " [5],\n",
       " [11],\n",
       " [11],\n",
       " [8],\n",
       " [8],\n",
       " [9],\n",
       " [10],\n",
       " [8],\n",
       " [10],\n",
       " [9],\n",
       " [8],\n",
       " [9],\n",
       " [7],\n",
       " [8],\n",
       " [9],\n",
       " [12],\n",
       " [6],\n",
       " [9],\n",
       " [8],\n",
       " [10],\n",
       " [6],\n",
       " [9],\n",
       " [6],\n",
       " [10],\n",
       " [9],\n",
       " [10],\n",
       " [8],\n",
       " [4],\n",
       " [8],\n",
       " [11],\n",
       " [8],\n",
       " [7],\n",
       " [9],\n",
       " [9],\n",
       " [9],\n",
       " [12],\n",
       " [7],\n",
       " [8],\n",
       " [7],\n",
       " [10],\n",
       " [10],\n",
       " [8],\n",
       " [7],\n",
       " [10],\n",
       " [5],\n",
       " [8],\n",
       " [4],\n",
       " [6],\n",
       " [9],\n",
       " [11],\n",
       " [3],\n",
       " [5],\n",
       " [8],\n",
       " [11],\n",
       " [8],\n",
       " [8],\n",
       " [8],\n",
       " [4],\n",
       " [7],\n",
       " [7],\n",
       " [9],\n",
       " [5],\n",
       " [7],\n",
       " [5],\n",
       " [9],\n",
       " [11],\n",
       " [10],\n",
       " [7],\n",
       " [6],\n",
       " [8],\n",
       " [7],\n",
       " [10],\n",
       " [10],\n",
       " [5],\n",
       " [9],\n",
       " [5],\n",
       " [9],\n",
       " [8],\n",
       " [10],\n",
       " [17],\n",
       " [6],\n",
       " [10],\n",
       " [8],\n",
       " [10],\n",
       " [12],\n",
       " [12],\n",
       " [7],\n",
       " [6],\n",
       " [6],\n",
       " [8],\n",
       " [8],\n",
       " [9],\n",
       " [8],\n",
       " [11],\n",
       " [8],\n",
       " [9],\n",
       " [10],\n",
       " [8],\n",
       " [7],\n",
       " [10],\n",
       " [10],\n",
       " [13],\n",
       " [6],\n",
       " [5],\n",
       " [10],\n",
       " [2],\n",
       " [11],\n",
       " [7],\n",
       " [6],\n",
       " [12],\n",
       " [10],\n",
       " [9],\n",
       " [9],\n",
       " [8],\n",
       " [9],\n",
       " [6],\n",
       " [10],\n",
       " [5],\n",
       " [6],\n",
       " [11],\n",
       " [9],\n",
       " [8],\n",
       " [6],\n",
       " [12],\n",
       " [9],\n",
       " [7],\n",
       " [13],\n",
       " [5],\n",
       " [9],\n",
       " [4],\n",
       " [9],\n",
       " [10],\n",
       " [13],\n",
       " [10],\n",
       " [12],\n",
       " [4],\n",
       " [7],\n",
       " [8],\n",
       " [8],\n",
       " [2],\n",
       " [5],\n",
       " [7],\n",
       " [8],\n",
       " [5],\n",
       " [7],\n",
       " [9],\n",
       " [10],\n",
       " [6],\n",
       " [6],\n",
       " [9],\n",
       " [6],\n",
       " [10],\n",
       " [11],\n",
       " [9],\n",
       " [5],\n",
       " [8],\n",
       " [6],\n",
       " [7],\n",
       " [11],\n",
       " [5],\n",
       " [8],\n",
       " [8],\n",
       " [10],\n",
       " [19],\n",
       " [8],\n",
       " [7],\n",
       " [5],\n",
       " [10],\n",
       " [9],\n",
       " [6],\n",
       " [14],\n",
       " [6],\n",
       " [9],\n",
       " [12],\n",
       " [11],\n",
       " [7],\n",
       " [9],\n",
       " [9],\n",
       " [10],\n",
       " [5],\n",
       " [10],\n",
       " [9],\n",
       " [6],\n",
       " [4],\n",
       " [16],\n",
       " [8],\n",
       " [8],\n",
       " [8],\n",
       " [5],\n",
       " [6],\n",
       " [8],\n",
       " [8],\n",
       " [10],\n",
       " [7],\n",
       " [7],\n",
       " [9],\n",
       " [9],\n",
       " [13],\n",
       " [6],\n",
       " [8],\n",
       " [8],\n",
       " [10],\n",
       " [6],\n",
       " [8],\n",
       " [8],\n",
       " [6],\n",
       " [12],\n",
       " [12],\n",
       " [4],\n",
       " [9],\n",
       " [8],\n",
       " [8],\n",
       " [6],\n",
       " [10],\n",
       " [11],\n",
       " [8],\n",
       " [8],\n",
       " [8],\n",
       " [8],\n",
       " [10],\n",
       " [9],\n",
       " [7],\n",
       " [6],\n",
       " [4],\n",
       " [9],\n",
       " [10],\n",
       " [8],\n",
       " [7],\n",
       " [7],\n",
       " [8],\n",
       " [17],\n",
       " [8],\n",
       " [10],\n",
       " [10],\n",
       " [10],\n",
       " [9],\n",
       " [3],\n",
       " [8],\n",
       " [6],\n",
       " [9],\n",
       " [8],\n",
       " [8],\n",
       " [10],\n",
       " [7],\n",
       " [6],\n",
       " [8],\n",
       " [10],\n",
       " [10],\n",
       " [8],\n",
       " [11],\n",
       " [8],\n",
       " [8],\n",
       " [11],\n",
       " [9],\n",
       " [10],\n",
       " [11],\n",
       " [6],\n",
       " [9],\n",
       " [4],\n",
       " [9],\n",
       " [7],\n",
       " [8],\n",
       " [6],\n",
       " [9],\n",
       " [8],\n",
       " [5],\n",
       " [7],\n",
       " [9],\n",
       " ...]"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "qtd_palavras =  Pipeline([\n",
    "                ('qtd_palavras', QtdPalavras())\n",
    "            ])\n",
    "qtd_palavras.fit_transform(dataset.noticia)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "** Feature Union com Pipelines**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "features = FeatureUnion([('text_process', text_process), \n",
    "                         ('qtd_palavras', qtd_palavras)\n",
    "                        ])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('text_process', Pipeline(memory=None,\n",
       "       steps=[('count', CountVectorizer(analyzer='word', binary=False, decode_error='strict',\n",
       "          dtype=<class 'numpy.int64'>, encoding='utf-8', input='content',\n",
       "          lowercase=True, max_df=1.0, max_features=None, min_df=1,\n",
       "          ngram_range=(1, 1), preprocessor=None, stop_words=None,\n",
       "          strip_accents=None, token_pattern='(?u)\\\\b\\\\w\\\\w+\\\\b',\n",
       "          tokenizer=None, vocabulary=None))])),\n",
       " ('qtd_palavras',\n",
       "  Pipeline(memory=None, steps=[('qtd_palavras', QtdPalavras())]))]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features.transformer_list"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Cria o pipeline que contém o objeto features**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "pip_features_1 = Pipeline([\n",
    "    ('features',features),\n",
    "    ('classifier', MultinomialNB()),\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Pipeline(memory=None,\n",
       "     steps=[('features', FeatureUnion(n_jobs=1,\n",
       "       transformer_list=[('text_process', Pipeline(memory=None,\n",
       "     steps=[('count', CountVectorizer(analyzer='word', binary=False, decode_error='strict',\n",
       "        dtype=<class 'numpy.int64'>, encoding='utf-8', input='content',\n",
       "        lowercase=True, max_d...nsformer_weights=None)), ('classifier', MultinomialNB(alpha=1.0, class_prior=None, fit_prior=True))])"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pip_features_1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "** Avalia modelo com quantidade de palavras**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predito     negativo   neutro   positivo   All\n",
      "Real                                          \n",
      " negativo        528       19        199   746\n",
      " neutro          104      111        237   452\n",
      " positivo        110       36        779   925\n",
      "All              742      166       1215  2123\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.6679393006205937"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "avalia_modelo(pip_features_1,dataset.noticia,dataset.sentimento)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Cria a classe Qtdcaracteres**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.base import BaseEstimator, TransformerMixin\n",
    "\n",
    "class QtdPalavras(BaseEstimator, TransformerMixin):\n",
    "    \n",
    "    def __init__(self):\n",
    "        pass\n",
    "    \n",
    "    def tamanho_noticia(self, texto):\n",
    "        return len(texto.split())\n",
    "    \n",
    "    def transform(self, news, y=None):\n",
    "        vet_tam = []\n",
    "        vet_tam_2 = []\n",
    "        for n in news:\n",
    "            vet_tam.append(self.tamanho_noticia(n))\n",
    "        vet_tam_2 = [[i] for i in vet_tam]\n",
    "        return vet_tam_2\n",
    "    \n",
    "    def fit(self, df, y=None):\n",
    "        return self\n",
    "    \n",
    "class QtdCaracteres(BaseEstimator, TransformerMixin):\n",
    "    \n",
    "    def __init__(self):\n",
    "        pass\n",
    "    \n",
    "    def conta_caracteres(self, texto):\n",
    "        return len(texto)\n",
    "    \n",
    "    def transform(self, news, y=None):\n",
    "        vet_tam = []\n",
    "        vet_tam_2 = []\n",
    "        for n in news:\n",
    "            vet_tam.append(self.conta_caracteres(n))\n",
    "        vet_tam_2 = [[i] for i in vet_tam]\n",
    "        return vet_tam_2\n",
    "    \n",
    "    def fit(self, df, y=None):\n",
    "        return self"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[[69],\n",
       " [61],\n",
       " [53],\n",
       " [66],\n",
       " [58],\n",
       " [56],\n",
       " [76],\n",
       " [51],\n",
       " [68],\n",
       " [53],\n",
       " [69],\n",
       " [97],\n",
       " [75],\n",
       " [72],\n",
       " [107],\n",
       " [46],\n",
       " [47],\n",
       " [69],\n",
       " [78],\n",
       " [40],\n",
       " [63],\n",
       " [77],\n",
       " [65],\n",
       " [46],\n",
       " [74],\n",
       " [48],\n",
       " [51],\n",
       " [74],\n",
       " [74],\n",
       " [36],\n",
       " [41],\n",
       " [58],\n",
       " [51],\n",
       " [59],\n",
       " [83],\n",
       " [56],\n",
       " [64],\n",
       " [69],\n",
       " [65],\n",
       " [82],\n",
       " [78],\n",
       " [69],\n",
       " [47],\n",
       " [45],\n",
       " [70],\n",
       " [54],\n",
       " [58],\n",
       " [66],\n",
       " [60],\n",
       " [42],\n",
       " [57],\n",
       " [58],\n",
       " [61],\n",
       " [66],\n",
       " [30],\n",
       " [117],\n",
       " [70],\n",
       " [79],\n",
       " [54],\n",
       " [64],\n",
       " [67],\n",
       " [47],\n",
       " [47],\n",
       " [27],\n",
       " [74],\n",
       " [25],\n",
       " [71],\n",
       " [59],\n",
       " [15],\n",
       " [56],\n",
       " [77],\n",
       " [50],\n",
       " [56],\n",
       " [56],\n",
       " [67],\n",
       " [55],\n",
       " [47],\n",
       " [68],\n",
       " [90],\n",
       " [71],\n",
       " [46],\n",
       " [74],\n",
       " [31],\n",
       " [93],\n",
       " [64],\n",
       " [48],\n",
       " [53],\n",
       " [65],\n",
       " [51],\n",
       " [56],\n",
       " [72],\n",
       " [84],\n",
       " [26],\n",
       " [69],\n",
       " [47],\n",
       " [56],\n",
       " [61],\n",
       " [43],\n",
       " [57],\n",
       " [79],\n",
       " [74],\n",
       " [57],\n",
       " [43],\n",
       " [31],\n",
       " [68],\n",
       " [52],\n",
       " [73],\n",
       " [47],\n",
       " [50],\n",
       " [43],\n",
       " [69],\n",
       " [105],\n",
       " [54],\n",
       " [57],\n",
       " [77],\n",
       " [48],\n",
       " [41],\n",
       " [22],\n",
       " [52],\n",
       " [41],\n",
       " [70],\n",
       " [56],\n",
       " [13],\n",
       " [55],\n",
       " [45],\n",
       " [56],\n",
       " [67],\n",
       " [56],\n",
       " [70],\n",
       " [57],\n",
       " [36],\n",
       " [54],\n",
       " [57],\n",
       " [53],\n",
       " [71],\n",
       " [70],\n",
       " [56],\n",
       " [34],\n",
       " [69],\n",
       " [52],\n",
       " [61],\n",
       " [84],\n",
       " [79],\n",
       " [32],\n",
       " [83],\n",
       " [57],\n",
       " [92],\n",
       " [62],\n",
       " [55],\n",
       " [70],\n",
       " [61],\n",
       " [83],\n",
       " [78],\n",
       " [54],\n",
       " [92],\n",
       " [75],\n",
       " [61],\n",
       " [72],\n",
       " [41],\n",
       " [62],\n",
       " [87],\n",
       " [53],\n",
       " [60],\n",
       " [66],\n",
       " [91],\n",
       " [46],\n",
       " [64],\n",
       " [43],\n",
       " [49],\n",
       " [90],\n",
       " [59],\n",
       " [58],\n",
       " [55],\n",
       " [80],\n",
       " [47],\n",
       " [52],\n",
       " [30],\n",
       " [58],\n",
       " [40],\n",
       " [25],\n",
       " [42],\n",
       " [68],\n",
       " [59],\n",
       " [56],\n",
       " [79],\n",
       " [50],\n",
       " [12],\n",
       " [44],\n",
       " [70],\n",
       " [57],\n",
       " [90],\n",
       " [78],\n",
       " [70],\n",
       " [34],\n",
       " [66],\n",
       " [51],\n",
       " [37],\n",
       " [84],\n",
       " [81],\n",
       " [89],\n",
       " [71],\n",
       " [42],\n",
       " [58],\n",
       " [70],\n",
       " [82],\n",
       " [66],\n",
       " [79],\n",
       " [59],\n",
       " [73],\n",
       " [72],\n",
       " [60],\n",
       " [57],\n",
       " [66],\n",
       " [42],\n",
       " [37],\n",
       " [55],\n",
       " [92],\n",
       " [59],\n",
       " [79],\n",
       " [68],\n",
       " [47],\n",
       " [66],\n",
       " [46],\n",
       " [62],\n",
       " [66],\n",
       " [41],\n",
       " [74],\n",
       " [76],\n",
       " [54],\n",
       " [75],\n",
       " [68],\n",
       " [51],\n",
       " [57],\n",
       " [80],\n",
       " [64],\n",
       " [43],\n",
       " [32],\n",
       " [56],\n",
       " [43],\n",
       " [56],\n",
       " [58],\n",
       " [59],\n",
       " [56],\n",
       " [44],\n",
       " [80],\n",
       " [75],\n",
       " [80],\n",
       " [50],\n",
       " [82],\n",
       " [55],\n",
       " [50],\n",
       " [70],\n",
       " [66],\n",
       " [41],\n",
       " [51],\n",
       " [39],\n",
       " [77],\n",
       " [50],\n",
       " [66],\n",
       " [39],\n",
       " [55],\n",
       " [75],\n",
       " [67],\n",
       " [68],\n",
       " [62],\n",
       " [44],\n",
       " [66],\n",
       " [60],\n",
       " [75],\n",
       " [63],\n",
       " [74],\n",
       " [50],\n",
       " [87],\n",
       " [52],\n",
       " [59],\n",
       " [55],\n",
       " [78],\n",
       " [72],\n",
       " [66],\n",
       " [66],\n",
       " [50],\n",
       " [47],\n",
       " [21],\n",
       " [72],\n",
       " [48],\n",
       " [67],\n",
       " [71],\n",
       " [41],\n",
       " [58],\n",
       " [33],\n",
       " [53],\n",
       " [57],\n",
       " [79],\n",
       " [102],\n",
       " [72],\n",
       " [70],\n",
       " [39],\n",
       " [78],\n",
       " [62],\n",
       " [70],\n",
       " [63],\n",
       " [49],\n",
       " [73],\n",
       " [65],\n",
       " [90],\n",
       " [61],\n",
       " [79],\n",
       " [72],\n",
       " [51],\n",
       " [70],\n",
       " [63],\n",
       " [34],\n",
       " [67],\n",
       " [66],\n",
       " [82],\n",
       " [68],\n",
       " [65],\n",
       " [33],\n",
       " [80],\n",
       " [55],\n",
       " [88],\n",
       " [63],\n",
       " [82],\n",
       " [72],\n",
       " [69],\n",
       " [69],\n",
       " [52],\n",
       " [67],\n",
       " [42],\n",
       " [76],\n",
       " [63],\n",
       " [82],\n",
       " [51],\n",
       " [72],\n",
       " [47],\n",
       " [71],\n",
       " [88],\n",
       " [33],\n",
       " [52],\n",
       " [38],\n",
       " [71],\n",
       " [38],\n",
       " [71],\n",
       " [79],\n",
       " [79],\n",
       " [40],\n",
       " [38],\n",
       " [41],\n",
       " [37],\n",
       " [52],\n",
       " [66],\n",
       " [61],\n",
       " [72],\n",
       " [29],\n",
       " [46],\n",
       " [61],\n",
       " [67],\n",
       " [71],\n",
       " [69],\n",
       " [47],\n",
       " [67],\n",
       " [96],\n",
       " [80],\n",
       " [72],\n",
       " [70],\n",
       " [63],\n",
       " [54],\n",
       " [42],\n",
       " [42],\n",
       " [70],\n",
       " [37],\n",
       " [51],\n",
       " [63],\n",
       " [40],\n",
       " [75],\n",
       " [40],\n",
       " [77],\n",
       " [64],\n",
       " [74],\n",
       " [56],\n",
       " [51],\n",
       " [74],\n",
       " [89],\n",
       " [36],\n",
       " [73],\n",
       " [64],\n",
       " [49],\n",
       " [65],\n",
       " [65],\n",
       " [68],\n",
       " [76],\n",
       " [59],\n",
       " [82],\n",
       " [38],\n",
       " [71],\n",
       " [61],\n",
       " [57],\n",
       " [49],\n",
       " [48],\n",
       " [69],\n",
       " [67],\n",
       " [67],\n",
       " [47],\n",
       " [42],\n",
       " [70],\n",
       " [69],\n",
       " [81],\n",
       " [86],\n",
       " [85],\n",
       " [71],\n",
       " [50],\n",
       " [60],\n",
       " [39],\n",
       " [57],\n",
       " [52],\n",
       " [134],\n",
       " [78],\n",
       " [74],\n",
       " [77],\n",
       " [37],\n",
       " [58],\n",
       " [109],\n",
       " [102],\n",
       " [76],\n",
       " [35],\n",
       " [81],\n",
       " [66],\n",
       " [67],\n",
       " [79],\n",
       " [90],\n",
       " [40],\n",
       " [89],\n",
       " [70],\n",
       " [51],\n",
       " [63],\n",
       " [46],\n",
       " [82],\n",
       " [78],\n",
       " [39],\n",
       " [32],\n",
       " [43],\n",
       " [67],\n",
       " [62],\n",
       " [60],\n",
       " [44],\n",
       " [75],\n",
       " [47],\n",
       " [47],\n",
       " [47],\n",
       " [74],\n",
       " [56],\n",
       " [50],\n",
       " [53],\n",
       " [53],\n",
       " [41],\n",
       " [42],\n",
       " [66],\n",
       " [53],\n",
       " [38],\n",
       " [43],\n",
       " [31],\n",
       " [31],\n",
       " [52],\n",
       " [74],\n",
       " [40],\n",
       " [79],\n",
       " [68],\n",
       " [73],\n",
       " [60],\n",
       " [66],\n",
       " [79],\n",
       " [77],\n",
       " [68],\n",
       " [66],\n",
       " [42],\n",
       " [71],\n",
       " [40],\n",
       " [54],\n",
       " [37],\n",
       " [66],\n",
       " [43],\n",
       " [43],\n",
       " [71],\n",
       " [79],\n",
       " [76],\n",
       " [41],\n",
       " [84],\n",
       " [45],\n",
       " [49],\n",
       " [71],\n",
       " [57],\n",
       " [90],\n",
       " [63],\n",
       " [74],\n",
       " [74],\n",
       " [31],\n",
       " [52],\n",
       " [80],\n",
       " [56],\n",
       " [61],\n",
       " [76],\n",
       " [81],\n",
       " [80],\n",
       " [48],\n",
       " [73],\n",
       " [93],\n",
       " [60],\n",
       " [47],\n",
       " [47],\n",
       " [58],\n",
       " [64],\n",
       " [42],\n",
       " [53],\n",
       " [73],\n",
       " [75],\n",
       " [76],\n",
       " [54],\n",
       " [70],\n",
       " [52],\n",
       " [78],\n",
       " [33],\n",
       " [51],\n",
       " [55],\n",
       " [64],\n",
       " [45],\n",
       " [82],\n",
       " [18],\n",
       " [64],\n",
       " [35],\n",
       " [71],\n",
       " [71],\n",
       " [80],\n",
       " [51],\n",
       " [65],\n",
       " [53],\n",
       " [95],\n",
       " [31],\n",
       " [48],\n",
       " [61],\n",
       " [57],\n",
       " [58],\n",
       " [37],\n",
       " [89],\n",
       " [51],\n",
       " [41],\n",
       " [38],\n",
       " [74],\n",
       " [15],\n",
       " [44],\n",
       " [64],\n",
       " [62],\n",
       " [85],\n",
       " [53],\n",
       " [51],\n",
       " [50],\n",
       " [38],\n",
       " [36],\n",
       " [43],\n",
       " [78],\n",
       " [65],\n",
       " [73],\n",
       " [72],\n",
       " [73],\n",
       " [78],\n",
       " [49],\n",
       " [59],\n",
       " [113],\n",
       " [42],\n",
       " [58],\n",
       " [55],\n",
       " [72],\n",
       " [66],\n",
       " [43],\n",
       " [84],\n",
       " [36],\n",
       " [49],\n",
       " [58],\n",
       " [56],\n",
       " [56],\n",
       " [80],\n",
       " [41],\n",
       " [87],\n",
       " [37],\n",
       " [44],\n",
       " [38],\n",
       " [65],\n",
       " [64],\n",
       " [45],\n",
       " [79],\n",
       " [69],\n",
       " [59],\n",
       " [73],\n",
       " [76],\n",
       " [29],\n",
       " [76],\n",
       " [52],\n",
       " [76],\n",
       " [73],\n",
       " [63],\n",
       " [46],\n",
       " [75],\n",
       " [66],\n",
       " [54],\n",
       " [49],\n",
       " [71],\n",
       " [162],\n",
       " [32],\n",
       " [73],\n",
       " [75],\n",
       " [54],\n",
       " [58],\n",
       " [80],\n",
       " [63],\n",
       " [92],\n",
       " [88],\n",
       " [32],\n",
       " [91],\n",
       " [66],\n",
       " [43],\n",
       " [63],\n",
       " [88],\n",
       " [78],\n",
       " [70],\n",
       " [47],\n",
       " [35],\n",
       " [36],\n",
       " [7],\n",
       " [85],\n",
       " [46],\n",
       " [74],\n",
       " [71],\n",
       " [47],\n",
       " [67],\n",
       " [53],\n",
       " [51],\n",
       " [88],\n",
       " [38],\n",
       " [67],\n",
       " [51],\n",
       " [79],\n",
       " [59],\n",
       " [83],\n",
       " [70],\n",
       " [88],\n",
       " [60],\n",
       " [74],\n",
       " [53],\n",
       " [64],\n",
       " [106],\n",
       " [88],\n",
       " [59],\n",
       " [81],\n",
       " [69],\n",
       " [60],\n",
       " [46],\n",
       " [68],\n",
       " [72],\n",
       " [50],\n",
       " [102],\n",
       " [55],\n",
       " [51],\n",
       " [62],\n",
       " [47],\n",
       " [45],\n",
       " [65],\n",
       " [60],\n",
       " [44],\n",
       " [53],\n",
       " [48],\n",
       " [75],\n",
       " [44],\n",
       " [63],\n",
       " [14],\n",
       " [54],\n",
       " [64],\n",
       " [65],\n",
       " [88],\n",
       " [77],\n",
       " [31],\n",
       " [70],\n",
       " [116],\n",
       " [66],\n",
       " [50],\n",
       " [102],\n",
       " [29],\n",
       " [50],\n",
       " [70],\n",
       " [59],\n",
       " [53],\n",
       " [72],\n",
       " [78],\n",
       " [65],\n",
       " [67],\n",
       " [65],\n",
       " [51],\n",
       " [75],\n",
       " [64],\n",
       " [85],\n",
       " [40],\n",
       " [73],\n",
       " [61],\n",
       " [92],\n",
       " [76],\n",
       " [65],\n",
       " [19],\n",
       " [81],\n",
       " [56],\n",
       " [69],\n",
       " [68],\n",
       " [67],\n",
       " [49],\n",
       " [99],\n",
       " [40],\n",
       " [77],\n",
       " [62],\n",
       " [91],\n",
       " [71],\n",
       " [57],\n",
       " [90],\n",
       " [63],\n",
       " [69],\n",
       " [60],\n",
       " [58],\n",
       " [58],\n",
       " [46],\n",
       " [44],\n",
       " [42],\n",
       " [40],\n",
       " [70],\n",
       " [69],\n",
       " [57],\n",
       " [24],\n",
       " [42],\n",
       " [83],\n",
       " [76],\n",
       " [68],\n",
       " [62],\n",
       " [55],\n",
       " [73],\n",
       " [69],\n",
       " [74],\n",
       " [65],\n",
       " [67],\n",
       " [54],\n",
       " [57],\n",
       " [56],\n",
       " [61],\n",
       " [87],\n",
       " [49],\n",
       " [60],\n",
       " [63],\n",
       " [64],\n",
       " [45],\n",
       " [71],\n",
       " [49],\n",
       " [67],\n",
       " [53],\n",
       " [72],\n",
       " [51],\n",
       " [28],\n",
       " [60],\n",
       " [88],\n",
       " [73],\n",
       " [60],\n",
       " [66],\n",
       " [59],\n",
       " [59],\n",
       " [74],\n",
       " [52],\n",
       " [56],\n",
       " [52],\n",
       " [84],\n",
       " [74],\n",
       " [57],\n",
       " [53],\n",
       " [76],\n",
       " [36],\n",
       " [58],\n",
       " [33],\n",
       " [46],\n",
       " [67],\n",
       " [86],\n",
       " [24],\n",
       " [45],\n",
       " [51],\n",
       " [71],\n",
       " [50],\n",
       " [81],\n",
       " [72],\n",
       " [35],\n",
       " [56],\n",
       " [60],\n",
       " [62],\n",
       " [28],\n",
       " [51],\n",
       " [41],\n",
       " [68],\n",
       " [69],\n",
       " [75],\n",
       " [60],\n",
       " [61],\n",
       " [52],\n",
       " [67],\n",
       " [76],\n",
       " [76],\n",
       " [43],\n",
       " [58],\n",
       " [43],\n",
       " [60],\n",
       " [61],\n",
       " [83],\n",
       " [126],\n",
       " [44],\n",
       " [72],\n",
       " [66],\n",
       " [69],\n",
       " [88],\n",
       " [85],\n",
       " [50],\n",
       " [38],\n",
       " [48],\n",
       " [65],\n",
       " [56],\n",
       " [65],\n",
       " [65],\n",
       " [91],\n",
       " [66],\n",
       " [60],\n",
       " [77],\n",
       " [68],\n",
       " [58],\n",
       " [75],\n",
       " [73],\n",
       " [99],\n",
       " [54],\n",
       " [35],\n",
       " [73],\n",
       " [15],\n",
       " [76],\n",
       " [41],\n",
       " [47],\n",
       " [78],\n",
       " [75],\n",
       " [68],\n",
       " [79],\n",
       " [57],\n",
       " [71],\n",
       " [52],\n",
       " [67],\n",
       " [40],\n",
       " [43],\n",
       " [80],\n",
       " [76],\n",
       " [55],\n",
       " [43],\n",
       " [68],\n",
       " [61],\n",
       " [53],\n",
       " [89],\n",
       " [43],\n",
       " [63],\n",
       " [38],\n",
       " [76],\n",
       " [82],\n",
       " [78],\n",
       " [74],\n",
       " [82],\n",
       " [28],\n",
       " [54],\n",
       " [49],\n",
       " [64],\n",
       " [14],\n",
       " [46],\n",
       " [48],\n",
       " [68],\n",
       " [37],\n",
       " [50],\n",
       " [75],\n",
       " [73],\n",
       " [52],\n",
       " [47],\n",
       " [67],\n",
       " [45],\n",
       " [79],\n",
       " [85],\n",
       " [74],\n",
       " [37],\n",
       " [58],\n",
       " [46],\n",
       " [53],\n",
       " [79],\n",
       " [40],\n",
       " [66],\n",
       " [58],\n",
       " [72],\n",
       " [143],\n",
       " [53],\n",
       " [53],\n",
       " [34],\n",
       " [57],\n",
       " [68],\n",
       " [47],\n",
       " [93],\n",
       " [48],\n",
       " [69],\n",
       " [80],\n",
       " [83],\n",
       " [55],\n",
       " [62],\n",
       " [58],\n",
       " [78],\n",
       " [44],\n",
       " [70],\n",
       " [62],\n",
       " [52],\n",
       " [38],\n",
       " [97],\n",
       " [59],\n",
       " [59],\n",
       " [60],\n",
       " [48],\n",
       " [48],\n",
       " [41],\n",
       " [64],\n",
       " [68],\n",
       " [56],\n",
       " [36],\n",
       " [61],\n",
       " [60],\n",
       " [105],\n",
       " [46],\n",
       " [58],\n",
       " [67],\n",
       " [70],\n",
       " [51],\n",
       " [76],\n",
       " [65],\n",
       " [49],\n",
       " [83],\n",
       " [91],\n",
       " [38],\n",
       " [79],\n",
       " [61],\n",
       " [56],\n",
       " [43],\n",
       " [75],\n",
       " [73],\n",
       " [64],\n",
       " [71],\n",
       " [62],\n",
       " [64],\n",
       " [72],\n",
       " [69],\n",
       " [42],\n",
       " [48],\n",
       " [32],\n",
       " [68],\n",
       " [62],\n",
       " [53],\n",
       " [46],\n",
       " [51],\n",
       " [49],\n",
       " [120],\n",
       " [60],\n",
       " [65],\n",
       " [72],\n",
       " [59],\n",
       " [67],\n",
       " [25],\n",
       " [67],\n",
       " [41],\n",
       " [68],\n",
       " [56],\n",
       " [55],\n",
       " [78],\n",
       " [55],\n",
       " [39],\n",
       " [56],\n",
       " [74],\n",
       " [90],\n",
       " [61],\n",
       " [65],\n",
       " [66],\n",
       " [65],\n",
       " [86],\n",
       " [61],\n",
       " [72],\n",
       " [87],\n",
       " [48],\n",
       " [70],\n",
       " [27],\n",
       " [60],\n",
       " [53],\n",
       " [69],\n",
       " [38],\n",
       " [64],\n",
       " [57],\n",
       " [37],\n",
       " [52],\n",
       " [57],\n",
       " ...]"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "qtd_caracteres =  Pipeline([\n",
    "                            ('qtd_caracteres', QtdCaracteres())\n",
    "                           ])\n",
    "qtd_caracteres.fit_transform(dataset.noticia)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**TF-IDF transformer**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import TfidfTransformer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<2123x3803 sparse matrix of type '<class 'numpy.float64'>'\n",
       "\twith 17325 stored elements in Compressed Sparse Row format>"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "text_process = Pipeline([\n",
    "                ('counts', CountVectorizer()),\n",
    "                ('tf-idf', TfidfTransformer())\n",
    "               ])\n",
    "text_process.fit_transform(dataset.noticia)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "features = FeatureUnion([('text_process', text_process), \n",
    "                         ('qtd_palavras', qtd_palavras),\n",
    "                         ('qtd_caracteres', qtd_caracteres),\n",
    "                     ])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "** Visualiza a lista de transformadores**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('text_process', Pipeline(memory=None,\n",
       "       steps=[('counts', CountVectorizer(analyzer='word', binary=False, decode_error='strict',\n",
       "          dtype=<class 'numpy.int64'>, encoding='utf-8', input='content',\n",
       "          lowercase=True, max_df=1.0, max_features=None, min_df=1,\n",
       "          ngram_range=(1, 1), preprocessor=None, stop_words=None,\n",
       "          strip_accents=None, token_pattern='(?u)\\\\b\\\\w\\\\w+\\\\b',\n",
       "          tokenizer=None, vocabulary=None)), ('tf-idf', TfidfTransformer(norm='l2', smooth_idf=True, sublinear_tf=False, use_idf=True))])),\n",
       " ('qtd_palavras',\n",
       "  Pipeline(memory=None, steps=[('qtd_palavras', QtdPalavras())])),\n",
       " ('qtd_caracteres',\n",
       "  Pipeline(memory=None, steps=[('qtd_caracteres', QtdCaracteres())]))]"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features.transformer_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "pip_features_3 = Pipeline([\n",
    "    ('features',features),\n",
    "    ('classifier', MultinomialNB()),\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predito     negativo   neutro   positivo   All\n",
      "Real                                          \n",
      " negativo        293        0        453   746\n",
      " neutro           53        5        394   452\n",
      " positivo         40        0        885   925\n",
      "All              386        5       1732  2123\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.5572387331759703"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "avalia_modelo(pip_features_3,dataset.noticia,dataset.sentimento)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Aplicando Pre-processadores em algumas features**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.preprocessing import StandardScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.30434783],\n",
       "       [0.30434783],\n",
       "       [0.30434783],\n",
       "       ...,\n",
       "       [0.39130435],\n",
       "       [0.13043478],\n",
       "       [0.43478261]])"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "qtd_palavras =  Pipeline([\n",
    "                ('qtd_palavras', QtdPalavras()),\n",
    "                ('Scaler', MinMaxScaler())\n",
    "            ])\n",
    "qtd_palavras.fit_transform(dataset.noticia)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 0.38583758],\n",
       "       [-0.06306401],\n",
       "       [-0.51196561],\n",
       "       ...,\n",
       "       [ 0.61028838],\n",
       "       [-1.07309261],\n",
       "       [ 1.84476778]])"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "qtd_caracteres =  Pipeline([\n",
    "                ('qtd_caracteres', QtdCaracteres()),\n",
    "                ('Scaler', StandardScaler())\n",
    "            ])\n",
    "qtd_caracteres.fit_transform(dataset.noticia)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "features = FeatureUnion([('text_process', text_process), \n",
    "                         ('qtd_palavras', qtd_palavras),\n",
    "                         ('qtd_caracteres', qtd_caracteres),\n",
    "                     ])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('text_process', Pipeline(memory=None,\n",
       "       steps=[('counts', CountVectorizer(analyzer='word', binary=False, decode_error='strict',\n",
       "          dtype=<class 'numpy.int64'>, encoding='utf-8', input='content',\n",
       "          lowercase=True, max_df=1.0, max_features=None, min_df=1,\n",
       "          ngram_range=(1, 1), preprocessor=None, stop_words=None,\n",
       "          strip_accents=None, token_pattern='(?u)\\\\b\\\\w\\\\w+\\\\b',\n",
       "          tokenizer=None, vocabulary=None)), ('tf-idf', TfidfTransformer(norm='l2', smooth_idf=True, sublinear_tf=False, use_idf=True))])),\n",
       " ('qtd_palavras', Pipeline(memory=None,\n",
       "       steps=[('qtd_palavras', QtdPalavras()), ('Scaler', MinMaxScaler(copy=True, feature_range=(0, 1)))])),\n",
       " ('qtd_caracteres', Pipeline(memory=None,\n",
       "       steps=[('qtd_caracteres', QtdCaracteres()), ('Scaler', StandardScaler(copy=True, with_mean=True, with_std=True))]))]"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features.transformer_list"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Importa o algoritmo SVM**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.svm import SVC"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "** Pipeline para Criação do Modelo usando SVM**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "pip_features_4 = Pipeline([\n",
    "    ('features',features),\n",
    "    ('classifier', SVC(kernel='linear')),\n",
    "])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Avaliando o modelo**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predito     negativo   neutro   positivo   All\n",
      "Real                                          \n",
      " negativo        536       49        161   746\n",
      " neutro          118      152        182   452\n",
      " positivo        122       63        740   925\n",
      "All              776      264       1083  2123\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.672638554970221"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "avalia_modelo(pip_features_4,dataset.noticia,dataset.sentimento)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}