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

Table 20

For K = 18 , training and testing samples variation for FTLRNN on testing data set for monthly sunspot time series and K = 20 for laser time series.

Time series		Sunspot time series 𝐾=18			Laser time series 𝐾=20
Training exemplars	Testing exemplars	MSE	NMSE	𝑟	MSE	NMSE	𝑟
10%	75%	0.01804	0.48780	0.74535	0.02676	0.80529	0.58788
20%	65%	0.01853	0.48428	0.75209	0.02361	0.68739	0.67208
30%	55%	0.01287	0.31169	0.82991	0.02569	0.79376	0.56198
40%	45%	0.01403	0.32187	0.83052	0.01659	0.66263	0.65482
50%	35%	0.02792	0.57914	0.73685	0.00703	0.30705	0.85539
60%	25%	0.02266	0.04075	0.7977	0.01635	0.543	0.71872
70%	15%	0.01885	0.39372	0.78804	0.03573	0.37992	0.37992
80%	05%	0.00745	0.21530	0.89769	0.00323	0.59234	0.59234