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

Table 18

For K = 6, training and testing samples variation for FTLRNN on testing data set for monthly sunspot time series and K = 5 for laser time series.

Time series		Sunspot time series 𝐾=6 Months			Laser time series 𝐾=5
Training exemplars	Testing exemplars	MSE	NMSE	𝑟	MSE	NMSE	𝑟
10%	75%	0.00707	0.19197	0.91703	0.00782	0.24068	0.87439
20%	65%	0.00451	0.11888	0.94153	0.00811	0.23969	0.88027
30%	55%	0.00423	0.10308	0.94712	0.00808	0.24814	0.87504
40%	45%	0.00438	0.10197	0.95081	0.00635	0.25307	0.87748
50%	35%	0.00654	0.13744	0.94252	0.00267	0.12898	0.94513
60%	25%	0.00554	0.1008	0.95528	0.00372	0.13558	0.93962
70%	15%	0.00473	0.10259	0.95359	0.00534	0.14838	0.92717
80%	05%	0.00499	0.15725	0.92195	0.01579	0.34297	0.83272