Skip to Main Content
Table 9:
Comparison of the performance of instance and feature selection methods in terms of the testing error and the storage requirements for datasets with many features and few instances. The table shows the win/loss record of each algorithm in the column against the algorithm in the row, the p-value of a two-tailed sign test on the win/loss record, ps, and the R+ and R values and p-value of the Wilcoxon test, pw. Significant differences at a confidence level of 95% are marked with a tick (✓).
Compared MethodMemetic IS+FS
Testing error   
 1-NN win/loss 9/9 
 ps 1.0000 
 R+/R 71.0/100.0 
 pw 0.5277 
 CHC IS+FS win/loss 11/7 
 ps 0.4807 
 R+/R 58.0/113.0 
 pw 0.2311 
 IMOEA win/loss 7/11 
 ps 0.4807 
 R+/R 103.0/68.0 
 pw 0.4460 
Storage requirements   
 CHC IS+FS win/loss 18/0 
 ps 0.0000 ✓ 
 R+/R 0.0/171.0 
 pw 0.0002 ✓ 
 IMOEA win/loss 18/0 
 ps 0.0065 ✓ 
 R+/R 0.0/171.0 
 pw 0.0002 ✓ 
Compared MethodMemetic IS+FS
Testing error   
 1-NN win/loss 9/9 
 ps 1.0000 
 R+/R 71.0/100.0 
 pw 0.5277 
 CHC IS+FS win/loss 11/7 
 ps 0.4807 
 R+/R 58.0/113.0 
 pw 0.2311 
 IMOEA win/loss 7/11 
 ps 0.4807 
 R+/R 103.0/68.0 
 pw 0.4460 
Storage requirements   
 CHC IS+FS win/loss 18/0 
 ps 0.0000 ✓ 
 R+/R 0.0/171.0 
 pw 0.0002 ✓ 
 IMOEA win/loss 18/0 
 ps 0.0065 ✓ 
 R+/R 0.0/171.0 
 pw 0.0002 ✓ 
Close Modal

or Create an Account

Close Modal
Close Modal