An Efficient and Effective Model to Handle Missing Data in Classification

Table 3

Accuracies achieved by simulation under MAR missing mechanism.

Variable	Model	Complete Variable	Missing Percent									Exclude Variable
			10%	20%	30%	40%	50%	60%	70%	80%	90%
	BART.m	66.56	66.26	65.96	65.63	65.37	65.08	64.79	64.52	64.24	64.01	63.65
	BART.i	66.56	66.04	65.47	64.76	64.00	63.28	62.42	61.69	61.17	60.64	63.65
	RF.i	63.63	63.25	62.81	62.31	61.68	61.11	60.47	59.91	59.40	59.07	60.74
	BART.m	66.56	66.53	66.51	66.54	66.54	66.57	66.60	66.61	66.65	66.67	66.67
	BART.i	66.56	66.50	66.45	66.31	66.14	65.81	65.42	64.78	63.88	62.17	66.67
	RF.i	63.65	63.55	63.48	63.29	63.17	62.90	62.51	61.96	61.34	59.87	65.91
	BART.m	66.56	66.22	65.87	65.48	65.08	64.68	64.30	63.87	63.48	63.13	62.62
	BART.i	66.56	65.98	64.90	63.74	62.64	61.52	60.34	59.31	58.27	56.98	62.62
	RF.i	63.63	63.17	62.36	61.59	60.81	59.95	59.01	58.13	57.25	55.97	63.03
	BART.m	66.56	66.30	66.08	65.71	65.27	64.76	63.99	63.25	62.28	61.32	60.14
	BART.i	66.56	66.14	65.50	64.55	63.42	62.06	60.39	58.71	56.77	54.81	60.14
	RF.i	63.63	63.35	62.90	62.27	61.49	60.63	59.32	57.95	56.21	54.46	59.60