How do I go about applying cost sensitive classifier in Orange?
I need to get a confusion matrix and predictions for a specified ratio of FP and FN. ROC curve widget lets me choose that ratio but where do I go from there?
I can't even get a reading on a threshold where performance line and ROC intersect. Also, I don't quite understand what prior target class probability is for. Model should be able to derive prior probabilities from the data set. Is it used for oversampling rare events?