Skip to content

Latest commit

 

History

History
37 lines (21 loc) · 1.07 KB

File metadata and controls

37 lines (21 loc) · 1.07 KB

home | copyright ©2016, tim@menzies.us

overview | syllabus | src | submit | chat


Feature Selection

Fire up the WEKA GUI using

    java -Xmx2048M -jar weka.jar &

Using the directions shown in class, for six data sets:

  • Perform feature selection using the J48 M trick;
  • Perform wrapper feature selection (on some smaller data sets);
  • Perform CFS feature subset selection (on some larger data sets).

Hand-in a table showing, for six data sets for (i) number of selected features, (ii) precision and (iii) recall numbers on the majority class in each data set:

  • Using all features;
  • Just using the selected features.