Concurrent, Performance-Based Methodology for Increasing the Accuracy and Certainty of Short-Term Neural Prediction Systems

Table 4

GNI forecasting performances of different approaches.

Method	Error type
	MAE	RMSE	MSE	MAPE
Single ANN	5.47	6.85	46.92	13.26
Combined ANN
Averaged	3.47	4.78	22.84	8.60
MAE weighted	3.00	3.70	13.68	7.47
RMSE weighted	2.95	3.20	12.27	7.39
MSE weighted	2.36	2.60	6.77	5.99
Combined ANN-LIN.REG.
MAE weighted	2.71	3.64	13.30	6.78
RMSE weighted	2.62	3.40	11.54	6.60
MSE weighted	2.24	2.51	6.29	5.73
ARIMA	13.26	23.12	24.71	8.60