Compound Autoregressive Network for Prediction of Multivariate Time Series

Table 2

Parameters of the network structure and training.

Datasets	Atmospheric quality data		Meteorological forecast data
Networks	Primary network	Auxiliary network	Primary network	Auxiliary network
Number of networks	1	5	1	3
Number of training times	1000	1000	1000	1000
Learning rate	0.01	0.01	0.02	0.02
Convergence error	0.002	0.002	0.001	0.001
Input delay	1 : 6	1 : 6	1 : 12	1 : 12
Output delay	1 : 6	1 : 6	1 : 12	1 : 12
Number of inputs	6	10	4	6
Number of outputs	1	10	1	6
Number of hidden-layer neurons	10	14	7	10