An Improved Generalized-Trend-Diffusion-Based Data Imputation for Steel Industry

Table 2

Imputation accuracies with different methods (hot blast stove’s BFG consumption).

Missing number	Methods	Group 1			Group 2			Group 3
		RMSE	NRMSE	MAPE	RMSE	NRMSE	MAPE	RMSE	NRMSE	MAPE
	Regression	57.16	0.0713	22.21	55.96	0.0777	28.56	35.56	0.0489	16.96
	EM	43.56	0.0543	17.88	45.82	0.0636	25.76	34.46	0.0474	17.58
3 points	SPLINE	36.72	0.0458	13.80	60.62	0.0842	31.69	22.10	0.0304	10.68
	GTD	9.13	0.0114	2.88	14.90	0.0207	5.85	14.50	0.0199	3.42
	iGTD	7.19	0.0090	2.05	14.42	0.0200	5.71	6.23	0.0086	2.87
	Regression	47.17	0.0578	23.07	29.10	0.0333	11.52	30.12	0.0356	11.56
	EM	26.15	0.0320	12.65	21.59	0.0247	10.96	26.44	0.0313	13.92
4 points	SPLINE	25.18	0.0308	7.80	21.06	0.0241	6.51	14.99	0.0177	5.25
	GTD	15.73	0.0193	4.74	11.78	0.0135	3.82	17.42	0.0206	5.85
	iGTD	15.09	0.0185	4.25	9.80	0.0112	3.37	16.63	0.0197	5.13
	Regression	28.84	0.0307	11.90	55.71	0.0612	30.84	39.92	0.0429	20.31
	EM	25.54	0.0272	13.35	39.99	0.0439	24.38	40.80	0.0439	25.78
5 points	SPLINE	47.33	0.0505	21.45	15.21	0.0167	7.00	33.42	0.0359	19.18
	GTD	20.12	0.0215	6.70	7.89	0.0087	3.93	6.67	0.0072	2.70
	iGTD	15.03	0.0160	5.97	7.09	0.0078	3.55	5.15	0.0055	2.16