Aula 3 Video 1 - Entendendo os dados e o problema



In [1]:

    
import pandas as pd
from sklearn.model_selection import train_test_split
from collections import Counter

Download Dataset: avaliacoes_usuario.csv



In [2]:

    
movies = pd.read_csv('datasets/avaliacoes_usuario.csv')



In [3]:

    
movies.head(5)









    Out[3]:







  
    
      
      Titulo
      Documentary
      Sci-Fi
      Mystery
      Horror
      Romance
      Thriller
      Crime
      Fantasy
      Comedy
      Animation
      Children
      Drama
      Adventure
      Duracao
      Investimento
      Gostou
    
  
  
    
      0
      Nixon (1995)
      0
      0
      0
      0
      0
      0
      0
      0
      0
      0
      0
      1
      0
      114.496547
      7.930748
      1
    
    
      1
      Leaving Las Vegas (1995)
      0
      0
      0
      0
      1
      0
      0
      0
      0
      0
      0
      1
      0
      110.140191
      18.276555
      1
    
    
      2
      Persuasion (1995)
      0
      0
      0
      0
      1
      0
      0
      0
      0
      0
      0
      1
      0
      105.747597
      16.582232
      1
    
    
      3
      Babe (1995)
      0
      0
      0
      0
      0
      0
      0
      0
      0
      0
      1
      1
      0
      126.131978
      13.004553
      1
    
    
      4
      Carrington (1995)
      0
      0
      0
      0
      1
      0
      0
      0
      0
      0
      0
      1
      0
      85.025469
      14.418120
      1



In [4]:

    
Counter(movies['Gostou'])









    Out[4]:





Counter({0: 135, 1: 543})



In [5]:

    
print('Gostou {}'.format((543/(135+543))))









    



Gostou 0.8008849557522124



In [6]:

    
print('Não gostou {}'.format((135/(135+543))))









    



Não gostou 0.19911504424778761



In [7]:

    
caract = movies[movies.columns[1:16]]
gostos = movies[movies.columns[16:]]



In [8]:

    
#treino, teste, treino_labels, teste_labels = train_test_split(caract, gostos, test_size=0.1)
treino, teste, treino_labels, teste_labels = train_test_split(caract, gostos)



In [9]:

    
type(treino)









    Out[9]:





pandas.core.frame.DataFrame



In [10]:

    
Counter(treino_labels['Gostou'])









    Out[10]:





Counter({0: 100, 1: 408})



In [11]:

    
Counter(teste_labels['Gostou'])









    Out[11]:





Counter({0: 35, 1: 135})



In [12]:

    
print('Gostou % {} do treino_label (deve manter a proporção +- 80%)'.format(404/(104+404)))









    



Gostou % 0.7952755905511811 do treino_label (deve manter a proporção +- 80%)



In [13]:

    
print('Gostou % {} do teste_label (deve manter a proporção +- 80%)'.format(139/(31+139)))









    



Gostou % 0.8176470588235294 do teste_label (deve manter a proporção +- 80%)

Aula 3 Video 2 - A regressão logística



In [14]:

    
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score
import numpy as np
from sklearn.naive_bayes import MultinomialNB,GaussianNB



In [15]:

    
type(treino)









    Out[15]:





pandas.core.frame.DataFrame



In [16]:

    
treino[0:1]









    Out[16]:







  
    
      
      Documentary
      Sci-Fi
      Mystery
      Horror
      Romance
      Thriller
      Crime
      Fantasy
      Comedy
      Animation
      Children
      Drama
      Adventure
      Duracao
      Investimento
    
  
  
    
      344
      0
      0
      0
      1
      0
      0
      0
      0
      0
      0
      0
      0
      0
      131.165837
      39.479492



In [17]:

    
treino = np.array(treino).reshape(len(treino), 15)
teste = np.array(teste).reshape(len(teste), 15)



In [18]:

    
type(treino)









    Out[18]:





numpy.ndarray



In [19]:

    
treino[0]









    Out[19]:





array([   0.        ,    0.        ,    0.        ,    1.        ,
          0.        ,    0.        ,    0.        ,    0.        ,
          0.        ,    0.        ,    0.        ,    0.        ,
          0.        ,  131.1658368 ,   39.47949166])



In [20]:

    
type(treino_labels)









    Out[20]:





pandas.core.frame.DataFrame



In [21]:

    
treino_labels.head(5)



In [22]:

    
treino_labels = treino_labels.values.ravel()
teste_labels = teste_labels.values.ravel()



In [23]:

    
type(treino_labels)









    Out[23]:





numpy.ndarray



In [24]:

    
treino_labels[0:5]









    Out[24]:





array([1, 1, 1, 1, 1])



In [25]:

    
treino_labels.shape









    Out[25]:





(508,)

Usando regressão logistica



In [26]:

    
modelo = LogisticRegression()
modelo.fit(treino, treino_labels)









    Out[26]:





LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,
          intercept_scaling=1, max_iter=100, multi_class='ovr', n_jobs=1,
          penalty='l2', random_state=None, solver='liblinear', tol=0.0001,
          verbose=0, warm_start=False)



In [27]:

    
previsoes = modelo.predict(teste)
previsoes









    Out[27]:





array([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
       1, 1, 1, 1, 1, 1, 1, 1, 1])



In [28]:

    
teste_labels.shape









    Out[28]:





(170,)



In [29]:

    
previsoes.shape









    Out[29]:





(170,)



In [30]:

    
acuracia = accuracy_score(teste_labels, previsoes)
acuracia









    Out[30]:





0.79411764705882348



In [31]:

    
zootopia = [0,0,0,0,0,0,0,1,1,1,1,0,1,110,27.74456356]
modelo.predict([zootopia])









    Out[31]:





array([1])

Usando Multinomial Naive Bayes



In [32]:

    
modelo_NB = MultinomialNB()



In [33]:

    
modelo_NB.fit(treino,treino_labels)









    Out[33]:





MultinomialNB(alpha=1.0, class_prior=None, fit_prior=True)



In [34]:

    
previsoes_NB = modelo_NB.predict(teste)



In [35]:

    
acuracia = accuracy_score(teste_labels, previsoes_NB)
acuracia









    Out[35]:





0.78235294117647058



In [36]:

    
zootopia = [0,0,0,0,0,0,0,1,1,1,1,0,1,110,27.74456356]
modelo_NB.predict([zootopia])









    Out[36]:





array([1])

EXTRA: Usando Gaussian Naive Bayes



In [37]:

    
modelo_GNB = GaussianNB()



In [38]:

    
modelo_GNB.fit(treino,treino_labels)









    Out[38]:





GaussianNB(priors=None)



In [39]:

    
previsoes_GNB = modelo_GNB.predict(teste)



In [40]:

    
acuracia = accuracy_score(teste_labels, previsoes_GNB)
acuracia









    Out[40]:





0.69411764705882351



In [41]:

    
zootopia = [0,0,0,0,0,0,0,1,1,1,1,0,1,110,27.74456356]
modelo_GNB.predict([zootopia])









    Out[41]:





array([1])

	Titulo	Romance	Children	Drama	Duracao	Investimento	Gostou
0	Nixon (1995)	0	0	1	114.496547	7.930748	1
1	Leaving Las Vegas (1995)	1	0	1	110.140191	18.276555	1
2	Persuasion (1995)	1	0	1	105.747597	16.582232	1
3	Babe (1995)	0	1	1	126.131978	13.004553	1
4	Carrington (1995)	1	0	1	85.025469	14.418120	1