def split_ds_train_test(supervised):
size_supervised = len(supervised)
global split_train_test
split_train_test = int(size_supervised * config.train_rate)
train, test = supervised[0:split_train_test], supervised[split_train_test:]
# transform the scale of the data
scaler, train_scaled, test_scaled = data_misc.scale(train, test)
x_train, y_train = train_scaled[:, 0:-1], train_scaled[:, -1]
x_test, y_test = test_scaled[:, 0:-1], test_scaled[:, -1]
return scaler, x_train, y_train, x_test, y_test
def split_ds_train_val_test(supervised):
size_supervised = len(supervised)
global split_train_val
split_train_val = int(size_supervised * config.train_val_rate)
global split_val_test
split_val_test = int(size_supervised * config.val_test_rate)
global split_train_test
split_train_test = split_train_val + split_val_test
train = supervised[0:split_train_val]
val = supervised[split_train_val:split_train_val + split_val_test]
test = supervised[split_train_val + split_val_test:]
# transform the scale of the data
scaler, train_scaled, val_scaled = data_misc.scale(train, val)
scaler, train_scaled, test_scaled = data_misc.scale(train, test)
#print(val[:, -1])
x_train, y_train = train_scaled[:, 0:-1], train_scaled[:, -1]
x_val, y_val = val_scaled[:, 0:-1], val_scaled[:, -1]
x_test, y_test = test_scaled[:, 0:-1], test_scaled[:, -1]
return scaler, x_train, y_train, x_val, y_val, x_test, y_test
df_trend[column] = series[column]
trend_supervised = supervised_diff_dt(df_trend, window_size_trend)
# we cut according to the biggest window size
trend_supervised = trend_supervised[total_window_size:, :]
# Concatenate with numpy
supervised = np.concatenate((trend_supervised, supervised), axis=1)
# Supervised reffers either avg_supervised or the combination of avg_supervised with bitcoin_values.
size_supervised = len(supervised)
split = int(size_supervised * 0.80)
train, test = supervised[0:split], supervised[split:]
# transform the scale of the data
scaler, train_scaled, test_scaled = data_misc.scale(train, test)
x_train, y_train = train_scaled[:, 0:-1], train_scaled[:, -1]
x_test, y_test = test_scaled[:, 0:-1], test_scaled[:, -1]
total_features = len(train[0])
print('Total Features: %i' % (total_features))
print('Total of supervised data: %i' % (size_supervised))
print(':: Train ::')
len_y_train = len(y_train)
# No Prediction
y_predicted_es = y_train
rmse, y_predicted = compare_train(len_y_train, y_predicted_es)
normal_train_results.append(rmse)
corr_normal_train_results.append(
data_misc.correlation(y_train, y_predicted))