Novel FTLRNN with Gamma Memory for Short-Term and Long-Term Predictions of Chaotic Time Series

Table 17

For K = 1, training and testing samples variation for FTLRNN on testing data set.

Time series		Sunspot time series			Laser time series 𝐾=1
Training exemplars	Testing exemplars	MSE	NMSE	𝑟	MSE	NMSE	𝑟
10%	75%	0.00320	0.08705	0.96121	0.00777	0.24076	0.87475
20%	65%	0.00258	0.6819	0.96605	0.01042	0.30799	0.84821
30%	55%	0.00188	0.04608	0.97791	0.01054	0.32361	0.84657
40%	45%	0.00218	0.05092	0.97928	0.00991	0.39119	0.81057
50%	35%	0.00273	0.05795	0.97837	0.00185	0.09233	0.98836
60%	25%	0.00227	0.04162	0.98163	0.00282	0.10597	0.94592
70%	15%	0.00336	0.07416	0.96799	0.00080	0.02293	0.98873
80%	05%	0.00153	0.05069	0.97512	0.00184	0.04229	0.97872