#!/usr/bin/python3
 import numpy as np
 from sklearn.preprocessing \
         import StandardScaler
 from keras.models import Sequential
 from keras.layers import Dense, Activation
 from keras.layers import LSTM
 import os

 os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'

 def window(npa, n=2):
     for start in range(npa.size-n+1):
         yield npa[start:start+n:1]

 input_size=3

 seq= np.array(
         [2,5,7,10,12]).astype('float64')

 print("learn input: " + str(seq))

 scaler = StandardScaler()
 seq = seq.reshape(-1,1)
 seq = scaler.fit_transform(seq)
 seq = seq.reshape(-1)

 X=np.array([])
 y=np.array([])

 for chunk in window(seq, n=input_size+1):
     X=np.append(X, chunk[:-1])
     y=np.append(y, chunk[-1])

 X=X.reshape((-1,input_size,1))
 y=y.reshape((-1,1))

 model = Sequential()
 model.add(LSTM(5,
     input_shape=(input_size,1)))
 model.add(Dense(1))
 model.add(Activation("linear"))
 model.compile(loss="mean_squared_error",
         optimizer="rmsprop")
 model.fit(X,y, epochs=500, verbose=0)

 print("\nresults:")
 for input in X:
     input=input.reshape(1,input_size,1)
     pred=model.predict(input)
     print(scaler.inverse_transform(
         input.reshape(-1,1)))
     print(scaler.inverse_transform(
         pred.reshape(-1,1)))

 test = seq[-input_size::1]
 print(scaler.inverse_transform(
     test.reshape(-1,1)))
 test=test.reshape(1,input_size,1)
 y1=model.predict(test)
 print(scaler.inverse_transform(
     y1.reshape(-1,1)))