CODEWITHSUNDEEP
machine-learningclassificationwekacross-validationresampling
manu
manu

Reputation: 107

Weka add more than one Meta filtered classifier

How can I add more than 1 meta filtered classifier? For example I want to apply SMOTE filter twice since my data has 3 classes with the ratio 6:3:1. So in order to balance the data I have to run the SMOTE thrice, by

first giving 200% --> then data will be 6:3:3

second giving 100% --> then data will be 6:6:3

third giving 100% agaim --? then data will be 6:6:6

I can do this in preprocess tab. But since i am using cross validation I have to use meta.filteredClassifier in classifiers tab.

How can I do this?

Upvotes: 0

Views: 595

Answers (1)

knb
knb

Reputation: 9303

I don't know how the SMOTE filter works; but it seems to me you want to apply multiple filters on one dataset, then use one classifier with cross-validation.

I am not sure if you are aware of the "MultiFilter" in the PreProcess Tab. The user interface is a bit weird. You have to click on ((1), see the top of my screenshot), then on (2, middle), then on (3). At (3) you can add several filters in a chain.

You can also configure the multiFilter from within the meta.FilteredClassifier window by clicking on the "filter" button in the classifier's dialog box.

enter image description here

Upvotes: 1

Related Questions