import numpy as np
from keras.models import Sequential
from keras.layers import Dense
from keras.utils.vis_utils import plot_model


# binary representation of the number
def binary_encode(i, num_digits):
    s = "{0:b}".format(i).zfill(num_digits)
    return np.array([int(x) for x in s])

# one hot encoding for even and odd labels
def one_hot_encode(i):
    if i % 2 == 0:
        return np.array([1, 0]) # even label
    else:
        return np.array([0, 1]) # odd label

NUM_DIGITS = 10
train_X = np.array([binary_encode(i, NUM_DIGITS) for i in range(101, 2 ** NUM_DIGITS)])
train_Y = np.array([one_hot_encode(i) for i in range(101, 2 ** NUM_DIGITS)])


# model
model = Sequential()
model.add(Dense(10, input_dim=10, activation="relu"))
model.add(Dense(5, activation="relu"))
model.add(Dense(2, activation="softmax"))
plot_model(model, to_file='model_plot.png', show_shapes=True, show_layer_names=True)
model.compile(loss='categorical_crossentropy', optimizer='adagrad', metrics=["accuracy"])
model.fit(train_X, train_Y, nb_epoch=20, verbose=1)


# one hot prediction to text
def prediction_to_text(i, prediction):
    return ["even", "odd"][prediction]


# try to predict even or odd for numbers from 1 to 100
numbers = np.arange(1, 101)
test_X = np.array([binary_encode(i, NUM_DIGITS) for i in numbers])
test_Y = model.predict_classes(test_X)
output = np.vectorize(prediction_to_text)(numbers, test_Y)
#print('predicted', output)


# correct answer
answer = np.array([])
for i in numbers:
    if i % 2 == 0:
        answer = np.append(answer, "even")
    else:
        answer = np.append(answer, "odd")
#print('correct', answer)


# evaluate
evaluate = np.array(answer == output)
print(np.count_nonzero(evaluate == True) / len(output))