In [2]:

    

from sklearn.datasets import fetch_20newsgroups
from keras.preprocessing.text import Tokenizer
from keras.preprocessing.sequence import pad_sequences
import keras
from keras.layers import Embedding, Dense, LSTM, GRU
from keras.models import Sequential


# In[33]:

from sklearn.model_selection import train_test_split, StratifiedShuffleSplit


# In[2]:

categories = ['comp.sys.mac.hardware',
 'comp.windows.x',
 'misc.forsale',
 'rec.autos',
'soc.religion.christian',
 'talk.politics.guns',
 'talk.politics.mideast']


# In[55]:

data = fetch_20newsgroups(shuffle=True, subset='train', categories=categories)


# In[54]:

test = fetch_20newsgroups(shuffle=True, subset='test', categories=categories)


# In[27]:

news, news_topics, _classes = data.data, data.target, data.target_names
# encoded_labels = [_classes.index(topic) for topic in news_topics]


# In[31]:




# In[35]:

# import string
# def translate_non_alphanumerics(to_translate, translate_to='_'):
#     not_letters_or_digits = string.punctuation #u'!"#%\'()*+,-./:;<=>?@[\]^_`{|}~'
#     translate_table = string.maketrans(not_letters_or_digits,
#                                        translate_to
#                                          *len(not_letters_or_digits))
#     translate_table = translate_table.decode("latin-1")
#     return to_translate.translate(translate_table)

# for i, item in enumerate(news):
#     news[i] = translate_non_alphanumerics(item)


# In[76]:

nb_words = 10000
tokenizer = Tokenizer(num_words=nb_words)
tokenizer.fit_on_texts(news)
sequences = Tokenizer.texts_to_sequences(tokenizer, news)


# In[81]:

word_index = tokenizer.word_index
print('Found %s unique tokens.' % len(word_index))


# In[32]:

max_seq_len = 1000
input_data = pad_sequences(sequences, maxlen=max_seq_len)
one_hot_labels = keras.utils.to_categorical(news_topics)

print('Shape of data tensor:', input_data.shape)
print('Shape of label tensor:', one_hot_labels.shape)


# In[107]:

Xtrain, Xvalid, ytrain, yvalid = train_test_split(input_data, one_hot_labels, test_size=0.2)


# In[94]:

len(word_index)


# In[113]:


    

    




Found 55199 unique tokens.
Shape of data tensor: (4059, 1000)
Shape of label tensor: (4059, 7)






    
Out[2]:





55199

In [3]:

    

embedding_vector_length = 64
model = Sequential()
model.add(Embedding(len(word_index), embedding_vector_length, input_length=max_seq_len, init='glorot_normal', 
                    W_regularizer=keras.regularizers.l2(0.01)))
model.add(LSTM(100, dropout_W=0.25))
model.add(Dense(7, activation='softmax'))


# In[114]:

model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])
print(model.summary())


# In[108]:

Xtrain.shape, ytrain.shape, Xvalid.shape, yvalid.shape


# In[ ]:


    

    




/Users/EH/.virtualenvs/ml/lib/python3.6/site-packages/ipykernel/__main__.py:4: UserWarning: Update your `Embedding` call to the Keras 2 API: `Embedding(55199, 64, input_length=1000, embeddings_initializer="glorot_normal", embeddings_regularizer=<keras.reg...)`
/Users/EH/.virtualenvs/ml/lib/python3.6/site-packages/ipykernel/__main__.py:5: UserWarning: Update your `LSTM` call to the Keras 2 API: `LSTM(100, dropout=0.25)`






    




_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
embedding_1 (Embedding)      (None, 1000, 64)          3532736   
_________________________________________________________________
lstm_1 (LSTM)                (None, 100)               66000     
_________________________________________________________________
dense_1 (Dense)              (None, 7)                 707       
=================================================================
Total params: 3,599,443.0
Trainable params: 3,599,443
Non-trainable params: 0.0
_________________________________________________________________
None






    
Out[3]:





((3247, 1000), (3247, 7), (812, 1000), (812, 7))

In [ ]:

    

model.fit(Xtrain, ytrain, epochs=20, batch_size=64)


    

    




Epoch 1/20
3247/3247 [==============================] - 1229s - loss: 0.5259 - acc: 0.8571    
Epoch 2/20
3247/3247 [==============================] - 597s - loss: 0.3972 - acc: 0.8572      
Epoch 3/20
 640/3247 [====>.........................] - ETA: 277s - loss: 0.4172 - acc: 0.8571

In [5]:

    

tes


    

    
Out[5]:





sklearn.datasets.base.Bunch

In [ ]: