Research Article

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

Table 6

Comparison of experimental results between CI_AMB and other methods (RMSE evaluation index of GBDT).

Original dataCI_AMBXGBoostLassoFCBF-MIC
Number of featuresRMSENumber of featuresRMSENumber of featuresRMSENumber of featuresRMSENumber of featuresRMSE

WYHXB798267.511580(19 + 61)232.735283269.164489255.966115265.0474
NYWZ10283258.4021220(59 + 161)234.8831212263.3908215256.217260265.2352
BlogData28022.724748(5 + 43)7.48224314.56604718.7933924.2629
RBuild103458.030235(16 + 19)417.144123458.278026466.85463461.7130
CCrime1270.106737(3 + 34)0.1091370.1176310.112150.1231
Average value201.3550178.4708201.1034199.5887203.2763