def eval(regressor, inputs):
real_stock_price = np.array(X)
predicted_stock_price = regressor.predict(inputs)
# rebuild the Structure
dataset_total = pd.DataFrame()
dataset_total['real'] = real_stock_price
dataset_total['predicted'] = predicted_stock_price
# real test data price VS. predicted price
predicted_stock_price = scaler.inverse_transform(dataset_total)
toler_rate = calc_diff(predicted_stock_price[:, 0], predicted_stock_price[:, 1])
err_cnt = error_count(predicted_stock_price[:, 0], predicted_stock_price[:, 1], toler_treshold = 3.0)
mse = mean_squared_error(predicted_stock_price[:, 0], predicted_stock_price[:, 1])
mape = mean_absolute_percentage_error(all_prices[:, 0], all_prices[:, 1])
predicted_stock_price_mod = create_pred_for_plot(predicted_stock_price)
print("pred_stock_prices_oneVector ", predicted_stock_price_mod, "size ",
predicted_stock_price_mod.shape, "\n")
dataset_test_total['predicted'] = predicted_stock_price_mod
print("Dataset_test_total ", dataset_test_total)
# real data price VS. predicted price
predicted_stock_price = scaler.inverse_transform(dataset_test_total)
# count of Wrong predicted value after applying treshold
err_cnt = error_count(predicted_stock_price[:, 0],
predicted_stock_price[:, 1],
toler_treshold=5.0)
# Calc difference between real data price and predicted price
diff_rate = calc_diff(predicted_stock_price[:, 0], predicted_stock_price[:, 1])
# show the inputs and predicted outputs
for i in range(len(predicted_stock_price[:, 1])):
print("X=%s, Predicted=%s" %
(predicted_stock_price[i, 1], predicted_stock_price[i, 0]))
print("Error count: ", err_cnt, "\n diff rate: ", diff_rate, "\n")
## Visualising the results
plot(predicted=predicted_stock_price[:, 1])
plot(real=predicted_stock_price[:, 0])
plot(predicted=predicted_stock_price[:, 1], real=predicted_stock_price[:, 0])
# MSE
mse = mean_squared_error(predicted_stock_price[:, 0], predicted_stock_price[:,
1])
print("MSE: ", mse)