Research Article

Prediction of Defective Software Modules Using Class Imbalance Learning

Table 6

Comparison on the basis of -measure values on eight datasets.

DatasetSVMCBNNNBRFLR-NNBBNC4.5LSTSVMWLSTSVM

CM10.19000.46640.42120.33980.31000.40140.48730.29750.54300.8191
KC10.27490.59080.37570.36370.49000.53410.40030.35120.85220.7639
PC10.60330.51720.41140.43650.33560.36060.60180.41020.68840.8019
PC30.62000.61400.33910.28280.33150.35340.39290.43950.51860.6371
PC40.60480.65420.45820.58230.50780.50260.55270.49760.62520.8236
MC20.54110.77930.44390.59520.41300.42370.46430.54640.75680.8318
KC20.63620.72690.52850.60180.64850.65160.68090.60220.84410.8478
KC30.36670.56920.48990.48590.36800.55970.61640.34430.63180.6762