esn = ESN(n_inputs = 1,
          n_outputs = 1,
          n_reservoir = 500,
          sparsity=0.2,
          random_state=23,
          spectral_radius=1.2,
          noise = 0.005)

trainlen = 1500

validation_set = []
for i in range(0,100):
    pred_training = esn.fit(np.ones(trainlen),amazon[i:trainlen+i])
    prediction = esn.predict(np.ones(2))
    validation_set.append(prediction[0])

def MSE(prediction, actual):
    return np.mean(np.power(np.subtract(np.array(prediction),actual),2))

def run_echo(sr, n, window):
    esn = ESN(n_inputs = 1,
          n_outputs = 1,
          n_reservoir = 500,
          sparsity=0.2,
          random_state=23,
          spectral_radius=sr,
          noise = n)

    trainlen = 1500
    current_set = []
    for i in range(0,100):
        pred_training = esn.fit(np.ones(trainlen),amazon[i:trainlen+i])
        prediction = esn.predict(np.ones(window))
        current_set.append(prediction[0])
    current_set = np.reshape(np.array(current_set),(-1,100))
    mse = MSE(current_set, amazon[trainlen:trainlen+100])

    return (mse, current_set)