0

I want to analyze my dataset with WEKA.

csv file

There are 60 apk files. Permission scores are calculated(1 point for neutral, 2-3-4 points for dangerous permissions) for each apk. I would like to create a threshold for (total dangerous permissions)/(total dangerous permissions + total safe permissions) with J48 using WEKA. I expect results such as "If the permission score is above %70, it is malware; vice versa. However, J48 option in Weka is greyed out. What should I change in this dataset?

Arthur Dent
  • 1
  • 1
  • What attribute did you select as class attribute? By default, the Weka Ecplorer uses the last attribute and your class seems to be the first one. J48 cannot handle numeric classes, only nominal ones. – fracpete Jun 18 '21 at 20:53

1 Answers1

1

The class attribute should be changed from numeric to nominal from Preprocess > Filter > Unsupervised > NumericToNominal, because J48 can handle Binary class, Missing class values, or Nominal class attributes, according to the documentation.

George
  • 37
  • 4