CODEWITHSUNDEEP
rmachine-learningweka
Reily Bourne
Reily Bourne

Reputation: 5297

Running Statistical tests in R from Decisions Trees by Weka

I am trying to figure out how to do this but Google does not seem to find me an answer.

I have a nice dataset that I am able to generate a pruned and unpruned decision trees in Weka. From this I can get the 10-fold cross-validation information which is nice.

But I would like to run statistical tests between the two decision trees, i.e. T-Test or Wilcoxon, using R. I have been suggested to use the DMwR and RWeka package but as I have no prior experience with this language, reading the RWeka docs and googling for tutorials or other explanations, I am coming up empty handed.

Upvotes: 1

Views: 313

Answers (1)

arutaku
arutaku

Reputation: 6087

As far as I know, you can run a t-test using WEKA's Experimenter. Where you select the dataset and the algorithms (they could be the same algorithm with different parameters) and then perform a t-test.

About Wilcoxon test, what I usually do is to "save" each model generated by WEKA (they would be Java Objects) and I read these objects in my Java program (maybe you can do it in R) and I perform the test in a Java program.

Upvotes: 1

Related Questions