A Novel Multilevel-SVD Method to Improve Multistep Ahead Forecasting in Traffic Accidents Domain

Table 3

Forecasting results for I-G2.

h; (week)				(%)			(%)
	MSVD + AR	SWT + AR	SWT + ANN	MSVD + AR	SWT + AR	SWT + ANN	MSVD + AR	SWT + AR	SWT + ANN
1	0.00048	0.62	0.91	99.9	97.9	97.1	99.9	98.9	98.5
2	0.00009	0.56	0.84	99.9	98.1	97.2	99.9	99.1	98.6
3	0.00132	0.85	1.64	99.9	97.2	94.4	99.9	98.6	97.2
4	0.00058	0.83	3.84	99.9	97.2	87.1	99.9	98.6	93.5
5	0.00105	1.12	8.68	99.9	96.3	71.4	99.9	98.2	85.7
6	0.00249	1.05	13.78	99.9	96.6	55.1	99.9	98.3	77.6
7	0.00761	1.22	14.67	99.9	95.9	47.2	99.9	97.9	73.6
8	0.02418	1.18	15.34	99.9	96.0	42.7	99.9	98.0	71.3
9	0.07071	10.4	—	99.8	65.6	—	99.9	82.8	—
10	0.19062	9.5	—	99.4	68.8	—	99.7	84.4	—
11	0.48291	12.5	—	98.5	58.6	—	99.2	79.3	—
12	1.15837	11.6	—	96.3	60.6	—	98.2	80.3	—
13	2.64161	19.6	—	91.6	36.4	—	95.8	68.2	—
14	5.77482	—	—	81.9	—	—	90.9	—	—
Min	0.00009	0.56	0.84	81.9	36.4	42.7	90.9	68.2	71.3
Max	5.77482	19.6	15.3	99.9	98.1	97.2	99.9	99.1	98.6
Mean 1–8	0.004	0.93	7.5	99.9	96.9	74.0	99.9	98.5	87.0
Mean 1–13	0.35	5.5	—	98.9	81.9	—	99.4	90.9	—
Mean 1–14	0.74	—	—	97.7	—	—	98.8	—	—