A novel intelligent approach for flight delay prediction

Journal of Big Data

Table 5 Accuracy of forecast horizons 2 h

Models	Traditional training model			Proposed FDPP-ML			Error reduction percent by FDPP-ML
Models	MAE	MSE	RMSE	MAE	MSE	RMSE	MAE (%)	MSE (%)	RMSE (%)
CATR	26.71	3544	59.53	17.34	2088	45.70	35	41	23
GBR	26.68	3616	60.14	18.24	2262	47.56	32	37	21
GRU	26.83	3620	60.17	18.57	2355	48.53	31	35	19
LGR	26.71	3570	59.75	17.59	2131	46.16	34	40	23
LR	27.22	3655	60.46	21.56	2733	52.28	21	25	14
LSTM	27.21	3671	60.59	19.64	2451	49.51	28	33	18
RFR	28.16	3685	60.71	17.29	2032	45.08	39	45	26
RNN	26.91	3665	60.54	19.08	2339	48.36	29	36	20
Stacking	26.29	3579	59.83	16.69	2071	45.50	37	42	24
Voting	26.87	3583	59.86	17.47	2111	45.95	35	41	23
Average error reduction percent by FDPP-ML							32	38	21