Stock Prediction Based on Optimized LSTM and GRU Models

Table 3

Learning results of PCA-LSTM.

Number of layers	Number of neurons	Look-back value	Train MSE	Train RMSE	Train MAE	Test MSE	Test RMSE	Test MAE	Train time
3	8	10	1690.9171	41.1208	28.1397	10971.7773	104.7463	72.8242	94.6602
3	8	20	2939.5332	54.2175	36.5242	42278.7188	205.6179	176.8455	198.6585
3	8	30	3896.0701	62.4185	50.7139	5464.5537	73.9226	54.9697	286.6041
3	8	40	2811.6401	53.0249	35.2591	37730.9688	194.2446	160.7494	410.7228
3	8	50	2484.0608	49.8404	38.5345	30902.8711	175.7921	154.2262	541.6164
2	8	10	1619.0308	40.2372	27.5215	19627.638	140.0987	116.2432	60.6061
2	8	20	2298.5891	47.9436	34.5403	17482.4219	132.2211	102.2909	128.5815
2	8	30	2054.7803	45.3297	32.6867	24208.5762	155.5911	125.7793	191.2626
2	8	40	2300.3635	47.9621	34.4281	27250.1621	165.0762	143.2569	252.3014
2	8	50	1614.6926	40.1832	26.5031	27625.6934	166.2098	140.6607	313.6309
3	16	10	1022.4178	31.9753	23.6002	19104.1875	138.2179	105.8687	278.2536
3	16	20	1744.8813	41.7718	31.4407	100281.7656	316.6730	261.6820	543.4168
3	16	30	932.1148	30.5306	23.0709	5806.1943	76.1984	58.5185	846.4980
3	16	40	1688.8053	41.0951	31.0483	72460.7422	269.1853	225.7833	1090.1818
3	16	50	1101.5414	33.1895	25.5879	4054.6895	63.6764	42.8904	1654.7813
2	16	10	755.3936	27.4844	19.6168	7201.1304	84.8595	50.9353	93.1576
2	16	20	642.6140	25.3498	18.2508	3843.0637	61.9924	48.0259	207.6672
2	16	30	734.1151	27.0946	19.6506	13394.8955	115.7363	89.6245	311.5668
2	16	40	737.1022	27.1496	19.1860	5955.7412	77.1734	59.1523	427.4087
2	16	50	941.9175	30.6907	21.1289	4154.1406	64.4526	50.9237	666.0416
3	32	10	599.8295	24.4914	19.4298	35741.8242	189.0551	124.2747	514.5233
3	32	20	594.4351	24.3810	18.6597	45484.7734	213.2716	169.4778	1064.2589
3	32	30	455.7460	21.3482	16.1783	17365.6328	131.7787	94.3915	1647.6020
3	32	40	805.4067	28.3797	23.0945	7770.9268	88.1529	58.4716	2789.5281
3	32	50	672.4887	25.9324	19.2775	91975.9453	303.2754	242.9042	2610.5209
2	32	10	495.8636	22.2680	16.6603	15152.0000	123.0935	90.1184	347.5115
2	32	20	627.7170	25.0543	18.4374	6696.0894	81.8296	50.5065	721.5676
2	32	30	454.6801	21.3232	15.7141	5454.8228	73.8568	51.7249	1089.6257
2	32	40	548.5927	23.4221	17.2341	17594.1758	132.6430	105.7155	1437.5246
2	32	50	407.0175	20.1747	14.6608	8100.0464	90.0003	66.1121	1808.6861

The number of epochs is 1000, learning rate is 0.001, and the activation function is .