03-24-2016 06:32 AM
I'm having a Credit Scoring problem.
I have a sample of Accepts and a sample of Rejects.
For the Accepts, I have a weight variable to adjust for oversampling (Y=1, Weight=1) (Y=0, Weight=30). The proportion of Y in my population is as follows: (Y=1 3%) (Y=0 97%).
As expected, I do not have a weight variable to account for oversampling in the Rejects dataset as I don't have a target. How would I go about building one?
And what is the difference between the weight that adjusts for oversampling and the weight needed to do Reject Inference?