Investigation of Bus Special Lane Performance Using Statistical Analysis and Optimization of the Signalized Intersection Delay by Machine Learning Methods

Table 17

Statistical results of the ANFIS model for predicting the delay at the signalized intersection.

Algorithm—input membership function	Training			Test			All data
	RMSE	COV	R^2	RMSE	COV	R^2	RMSE	COV	R^2
2-trimf	4.1322	7.8626	0.9852	28.1477	48.6353	0.8005	11.9697	22.3637	0.9450
3-trimf	0.0044	0.0084	0.9904	48.5948	83.9649	0.3420	20.7209	38.7141	0.8499
4-trimf	0.0003	0.0006	0.9909	37.5285	64.8442	0.6524	15.1418	28.2903	0.9174
2-trapmf	6.8831	13.0967	0.9747	35.1272	61.3576	0.6419	16.4082	31.2607	0.8987
3-trapmf	0.0028	0.0054	0.9904	24.4196	42.6543	0.8222	10.5330	20.0671	0.9529
4-trapmf	1.8748	3.5674	0.9898	37.9973	66.3708	0.5825	16.4398	31.3208	0.8983
2-gbellmf	4.3868	8.3469	0.9844	34.0760	59.5214	0.6625	15.2360	28.4664	0.9165
3-gbellmf	0.0050	0.0095	0.9901	36.3502	63.4936	0.6172	15.6790	29.2940	0.9120
4-gbellmf	0.0004	0.0007	0.9908	32.0985	56.0671	0.6994	13.8451	26.3774	0.9253
2-gaussmf	4.1741	7.9423	0.9851	68.6383	119.8921	0.3420	29.8511	55.7729	0.7048
3-gaussmf	0.0064	0.0123	0.9910	20.1961	35.2771	0.8755	8.7112	16.2757	0.9666
4-gaussmf	0.0005	0.0009	0.9910	50.2771	87.8203	0.2758	21.6861	41.3160	0.8299
2-gauss2mf	4.7970	9.1275	0.9831	44.8612	78.3600	0.4216	19.8416	37.0713	0.8645
3-gauss2mf	0.0058	0.0109	0.9907	8.8807	15.5121	0.9887	3.0153	4.3401	0.9905
4-gauss2mf	0.0011	0.0022	0.9908	28.2451	49.3365	0.7653	12.1830	23.2109	0.9402