A novel intelligent approach for flight delay prediction

Journal of Big Data

Table 3 Accuracy of forecast horizons 12 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	30.62	4341	65.88	25.87	3483	59.01	16	20	10
GBR	31.02	4422	66.50	26.48	3777	61.45	15	15	8
GRU	31.04	4441	66.64	26.46	3966	62.98	15	11	5
LGR	30.87	4366	66.08	25.96	3574	59.79	16	18	10
LR	31.25	4485	66.97	29.78	4329	65.80	5	3	2
LSTM	31.25	4467	66.84	27.02	4180	64.66	14	6	3
RFR	31.76	4477	67.27	26.30	3501	59.17	17	22	12
RNN	31.17	4488	67.00	26.78	3999	63.24	14	11	6
Stacking	30.34	4385	66.22	24.82	3541	59.50	18	19	10
Voting	30.87	4392	66.28	25.76	3549	59.57	17	19	10
Average error reduction percent by FDPP-ML							15	14	8