import matplotlib.pyplot as plt
from keras import models
from keras import layers
from keras.datasets import imdb
import numpy as np

(train_data, train_labels),(test_data, test_labels) = 
    imdb.load_data(num_words=10000)             //1
print(train_data)

reverse_word_index = 
dict([(value, key) for (key, value) in word_index.items()]) //2

decoded_review = ' '.join([reverse_word_index.get(i - 3, '?') for i in train_data[0]])  //3
print("--------------")
print(decoded_review)
print("--------------")

#3 Break the data into training and testing data
def vectorize_sequences(sequences, dimension=10000):
    results = np.zeros((len(sequences), dimension))
    for i, sequence in enumerate(sequences):       
        results[i, sequence] = 1
    return results
x_train = vectorize_sequences(train_data)           #Convert traning data into array of 0 & 1 of size=10k
x_test = vectorize_sequences(test_data)             #Convert test data into array of 0 & 1 of size=10k
y_train = np.asarray(train_labels).astype('float32')
y_test = np.asarray(test_labels).astype('float32')
print('============')
print(x_train)
print('!!!!!!!!!!!!')
print(x_test)
print('!!!!!!!!!!!!')
print(y_train)                                      #[0. 1. 1. ... 0. 0. 0.]
print('!!!!!!!!!!!!')
print(y_test)                                       #[0. 1. 1. ... 0. 0. 0.]
print('============')

# 4. Create Neural Network
model = models.Sequential()
model.add(layers.Dense(16, activation='relu', input_shape=(10000,)))
model.add(layers.Dense(16, activation='relu'))
model.add(layers.Dense(1, activation='sigmoid'))

# 5. Loss function, Optimizer
model.compile(optimizer='rmsprop',loss='binary_crossentropy',metrics=['accuracy'])

# 6. Create Validation set(10k samples) from training data
x_val = x_train[:10000]             #x_train[] is matrix to be inputted into NN
partial_x_train = x_train[10000:]
y_val = y_train[:10000]
partial_y_train = y_train[10000:]

# 7. Train the model
history = model.fit(partial_x_train,partial_y_train,
        epochs=20,batch_size=512,
        validation_data=(x_val, y_val))
print('Traning Complete')
print(history.history.keys())       #dict_keys(['val_loss', 'val_accuracy', 'loss', 'accuracy'])

# 8a. Plot training vs Validation loss
history_dict = history.history
acc = history_dict['accuracy']
loss_values = history_dict['loss']
val_loss_values = history_dict['val_loss']
epochs = range(1, len(acc) + 1)
plt.plot(epochs, loss_values, 'bo', label='Training loss')
plt.plot(epochs, val_loss_values, 'b', label='Validation loss')
plt.title('Training and validation loss')
plt.xlabel('Epochs')
plt.ylabel('Loss')
plt.legend()
plt.show()

# 8b. PLOTTING TRANING vs VALIDATION_ACCURACY
plt.clf()
acc_values = history_dict['accuracy']
val_acc_values = history_dict['val_accuracy']
plt.plot(epochs, acc, 'bo', label='Training acc')
plt.plot(epochs, val_acc_values, 'b', label='Validation acc')
plt.title('Training and validation accuracy')
plt.xlabel('Epochs')
plt.ylabel('Loss')
plt.legend()
plt.show()

# 9. Predict the output
print(model.evaluate(partial_x_train,partial_y_train))
            

Layer   hidden-Units    Activation
  1      16             relu(rectified linear unit)
  2      16             relu
  3      1              sigmoid
          

    input nD tensor
            \/
   -----------------------------------------
  |Layer-1: Dense(units=512)(activation=relu)|
   -----------------------------------------
            \/ nD tensor
   -----------------------------------------
  |Layer-2: Dense(units=512)(activation=relu)|<------
   -----------------------------------------        |
            \/ nD tensor                            |
   -----------------------------------------        |
  |Layer-3 Dense(units=1)(activation=sigmoid)|      |
   -----------------------------------------        |
            \/                                      |
        output(probability)                         |
            \/                                      |
    |loss function|                                 |
            \/                                      |
            loss score  ------|optimizer|-----------