import numpy as np
from keras.models import Sequential
from keras.layers.core import Dense, Dropout, Activation
from keras.optimizers import SGD, Adam, RMSprop
from keras.utils import np_utils
from time import time
from pickle import dump
trainfile = "train.csv"
t = time()
data = np.loadtxt(open(trainfile,'r'),delimiter=',',skiprows=1) #http://softwarerecs.stackexchange.com/questions/7463/fastest-python-library-to-read-a-csv-file
y,X =  data[:,0:1],data[:,1:]/255
print("Loading :",time()-t," sec")
#check shape
print(X,y)

outname = "modell"
dropout_rate = 0.4
batch_size = 128
nb_classes = 10
nb_epoch = 20
#split train|valid
#if (False):
#	split_len = 2*len(X)//3
#	X_train,X_test = X[:split_len],X[split_len:]
#	y_train,y_test = y[:split_len].astype('int32'),y[split_len:].astype('int32')
#	y_train,y_test = np_utils.to_categorical(y_train,nb_classes),np_utils.to_categorical(y_test,nb_classes)
#	print(y_train,y_test)
#else:
X_train = X
y_train = np_utils.to_categorical(y.astype('int32'),nb_classes)


#build nn
model = Sequential()
model.add(Dense(512, input_shape=(784,)))
model.add(Activation('relu'))
model.add(Dropout(dropout_rate))
model.add(Dense(512))
model.add(Activation('relu'))
model.add(Dropout(dropout_rate))
model.add(Dense(10))
model.add(Activation('softmax'))
model.compile(loss='categorical_crossentropy',
              optimizer=RMSprop(),
              metrics=['accuracy'])
#model.summary()
history = model.fit(X_train, y_train,
                    batch_size=batch_size, nb_epoch=nb_epoch,
                    verbose=1)
with open(outname+".json", "w") as json_file:
    json_file.write(model.to_json())
model.save_weights(outname+".h5")