The Start of Combustion Prediction for Methane-Fueled HCCI Engines: Traditional vs. Machine Learning Methods

Table 5

The performance of employed models on the test dataset.

Parameter	Model
	1	2	3	4	5	6	7	8	9	10	11
Residual less than 2 CAD (%)	98.81423	92.49012	98.81423	0.395257	98.81423	9.486166	98.41897	26.08696	44.26877	77.0751	15.41502
Residual less than 3 CAD (%)	100	99.60474	100	0.790514	100	16.60079	100	34.78261	55.73123	89.72332	26.48221
Residual less than 4 CAD (%)	100	100	100	0.790514	100	25.29644	100	46.24506	73.12253	95.25692	33.99209
MAE	0.8233	0.9926	0.7878	484.0682	0.8226	7.8129	0.7766	4.3073	2.7197	1.3818	7.7372
MSE	0.9541	1.4513	0.8547	352987.6	0.9519	87.2833	0.8889	25.8072	11.5298	3.4353	92.1935
RMSE	0.9768	1.2047	0.9245	594.1276	0.9756	9.3425	0.9428	5.08	3.3955	1.8534	9.6017
Score	0.9039	0.8539	0.9139	-35526.2	0.9041	−7.7848	0.9105	−1.5974	−0.1604	0.6542	−8.279