Desktop productivity for business analysts and programmers

How to adjust TPR / TNR for oversampled data?

Reply
New Contributor
Posts: 4

How to adjust TPR / TNR for oversampled data?

My original dataset has a proportion of Y = 90% / N = 10%. Using the Sample node, this was made 50% / 50% before running through a Decision Tree node. I understand the Decisions node allows for adjustment of Profit based on prior probabilities, however I read that classification rates are not adjusted. 

 

How do I adjust my True Positive Rate / True Negative Rate in this scenario given the oversampling and prior probabilities?

Frequent Contributor
Posts: 136

Re: How to adjust TPR / TNR for oversampled data?

adjacent to the decision weights tab in the same data source step is a target priors tab for just that.
Ask a Question
Discussion stats
  • 1 reply
  • 317 views
  • 0 likes
  • 2 in conversation