Table 5 Features selection test according to feature selection methods
From: Objective and automatic classification of Parkinson disease with Leap Motion controller
Test number | Method | Selects | Search algorithms | Selected subsets/features (mean value among the three trials) |
|---|---|---|---|---|
1 | Principal components | Attributes | Ranker | Num-OC, Wcl, tetaSD-OC, Wop, Num-PS |
2 | SVM | Attributes | Ranker | fSD_PS, WcTF, tetaSD-PS, tetaSD-OC, Wps, WoTF, Num-PS |
3 | Consistency | Subset | Greedy SW | Num-OC, Wop, fSD-PS, tetaSD-PS |
4 | J48 | Subset | Greedy SW | Num-OC, WcTF, Wop |
5 | Filtered subset evaluation | Subset | Genetic search | Num-PS, Num-OC, tetaSD-PS, fSD-PS, Wcl |
6 | Information gain | Attributes | Ranker | PwrP, fSD-TF, Num-OC, tetaSD-TF, Wsp, Wps, Num-PS |
7 | Gain ratio | Attributes | Ranker | PwrP, fSD-TF, Num-OC, tetaSD-TF, Wsp, Wps, Num-PS |
8 | Chi square attribute evaluation | Attributes | Ranker | PwrP, fSD-TF, tetaSD-TF, Num-OC, Wsp, Wps, Num-PS |