Hybrid Model for Method for Short-Term Traffic Flow Prediction Based on Secondary Decomposition Technique and ELM

Table 3

The MAE value of the involved models by cross-validation strategy.

Iterations	ARIMA	DBN	ELM	Model1	Model2	Model3	Proposed
1	4.0617	4.1237	4.0108	2.6818	0.7269	0.6729	0.6246
2	4.0618	4.1854	3.9812	2.6902	0.6157	0.6632	0.6973
3	4.0617	4.1045	3.9401	2.6229	0.5979	0.6690	0.6107
4	4.0594	4.1206	4.0262	2.6691	0.6402	0.6696	0.6127
5	4.0474	4.1304	4.0089	2.6742	0.7839	0.6621	0.6254
6	4.0617	4.0853	4.0454	2.7429	0.6292	0.6695	0.6451
7	4.0520	4.1180	4.0259	2.6335	0.5874	0.6708	0.6325
8	4.0618	4.4703	3.9819	2.7134	0.6090	0.6598	0.6675
9	4.0454	4.2452	3.9766	2.6952	0.6974	0.6690	0.6498
10	4.0618	4.1636	3.9850	2.6780	0.6251	0.6617	0.6634
11	4.0616	4.1455	4.0277	2.6868	0.7910	0.6706	0.6036
12	4.0618	4.1459	4.0456	2.6919	0.6416	0.6651	0.6775
13	4.0613	4.1520	4.0169	2.5857	0.5978	0.6628	0.5886
14	4.0648	4.1253	4.0906	2.7262	0.5948	0.6826	0.7869
15	4.0594	4.1019	4.0371	2.6931	0.6642	0.6494	0.5967
16	4.0614	4.1913	3.9869	2.6964	0.6378	0.6613	0.5902
17	4.0462	4.1414	3.9815	2.6349	0.6835	0.6596	0.6482
18	4.0617	4.1169	4.0321	2.6755	0.6652	0.6605	0.6534
19	4.0616	4.1329	4.0248	2.7189	0.7166	0.6575	0.6020
20	4.0616	4.1700	3.9875	2.6862	0.6363	0.6577	0.7142
Average	4.0588	4.1585	4.0106	2.6798	0.6571	0.6647	0.6445