Research on Hybrid Feature Selection Method Based on Iterative Approximation Markov Blanket

Table 7

Comparison of experimental results of CI_AMB with other methods (RMSE evaluation index of XGBoost).

	Original data		CI_AMB		XGBoost		Lasso		FCBF-MIC
	Number of features	RMSE	Number of features	RMSE	Number of features	RMSE	Number of features	RMSE	Number of features	RMSE
WYHXB	798	227.9061	80(19 + 61)	205.0669	83	221.8774	89	214.0560	15	229.7367
NYWZ	10283	219.7160	220(59 + 161)	201.5748	212	220.3312	215	225.1712	60	225.1525
BlogData	280	8.6356	48(5 + 43)	4.1587	43	10.0949	47	10.2909	9	10.8045
RBuild	103	264.5195	35(16 + 19)	255.1114	23	269.8928	26	261.3095	3	278.6242
CCrime	127	0.1447	37(3 + 34)	0.1443	37	0.1487	31	0.1483	5	0.1492
Average value		144.1844		133.2112		144.4690		142.1952		148.8934