Chapter 4: Introduction to convolutional networks

Activity 5: Sentiment Analysis on a real-life dataset

Solution:

1. Import the necessary classes

   from keras.preprocessing.text import Tokenizer

   from keras.models import Sequential

   from keras import layers

   from keras.preprocessing.sequence import pad_sequences

   import numpy as np

   import pandas as pd

2. Define your variables and parameters.

   epochs = 20

   maxlen = 100

   embedding_dim = 50

   num_filters = 64

   kernel_size = 5

   batch_size = 32

3. Import the data.

   data = pd.read_csv('data/sentiment labelled sentences/yelp_labelled.txt',names=['sentence', 'label'], sep='\t')

   data.head()

   Printing this out on a Jupyter notebook should display:

   

   Figure 4.27: Labelled dataset

4. Select the 'sentence' and 'label' columns

   sentences=data['sentence'].values

   labels=data['label'].values

5. Split your data into training and test set

   from sklearn.model_selection import train_test_split

   X_train, X_test, y_train, y_test = train_test_split(

   sentences, labels, test_size=0.30, random_state=1000)

6. Tokenize

   tokenizer = Tokenizer(num_words=5000)

   tokenizer.fit_on_texts(X_train)

   X_train = tokenizer.texts_to_sequences(X_train)

   X_test = tokenizer.texts_to_sequences(X_test)

   vocab_size = len(tokenizer.word_index) + 1 #The vocabulary size has an additional 1 due to the 0 reserved index

7. Pad in order to ensure that all sequences have the same length

   X_train = pad_sequences(X_train, padding='post', maxlen=maxlen)

   X_test = pad_sequences(X_test, padding='post', maxlen=maxlen)

8. Create the model. Note that we use a sigmoid activation function on the last layer and the binary cross entropy for calculating loss. This is because we are doing a binary classification.

   model = Sequential()

   model.add(layers.Embedding(vocab_size, embedding_dim, input_length=maxlen))

   model.add(layers.Conv1D(num_filters, kernel_size, activation='relu'))

   model.add(layers.GlobalMaxPooling1D())

   model.add(layers.Dense(10, activation='relu'))

   model.add(layers.Dense(1, activation='sigmoid'))

   model.compile(optimizer='adam',

   loss='binary_crossentropy',

   metrics=['accuracy'])

   model.summary()

   The above code should yield

   

   Figure 4.28: Model summary

   The model can be visualized as follows as well:

   

   Figure 4.29: Model visualization

9. Train and test the model.

   model.fit(X_train, y_train,

   epochs=epochs,

   verbose=False,

   validation_data=(X_test, y_test),

   batch_size=batch_size)

   loss, accuracy = model.evaluate(X_train, y_train, verbose=False)

   print("Training Accuracy: {:.4f}".format(accuracy))

   loss, accuracy = model.evaluate(X_test, y_test, verbose=False)

   print("Testing Accuracy: {:.4f}".format(accuracy))

   The accuracy output should be as follows:

Figure 4.30: Accuracy score