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

Table 16

For K = 100, training and testing samples variation for FTLRNN on testing data set.

Time series		Mackey-Glass time series			Duffing time series
Training exemplars	Testing exemplars	MSE	NMSE	𝑟	MSE	NMSE	𝑟
10%	75%	0.029161	0.47623	0.7294	0.01359	0.54149	0.80092
20%	65%	0.02897	0.45601	0.74595	0.00927	0.07219	0.96339
30%	55%	0.03182	0.48156	0.72938	0.00116	0.07810	0.962300
40%	45%	0.03439	0.50398	0.71524	0.00875	0.05663	0.97196
50%	35%	0.04027	0.54749	0.68781	0.00883	0.06062	0.96947
60%	25%	0.05345	0.65267	0.62265	0.00117	0.07870	0.96000
70%	15%	0.08017	0.81243	0.53108	0.01033	0.06935	0.96570
80%	05%	0.21006	1.47654	0.24862	0.01141	0.07715	0.96069