03-14-2016 01:43 PM - edited 03-14-2016 01:44 PM
I have a dataset for a specific time period(say May 2015). and I have another dataset(say Aug 2015). I built a logistic regression model on the May data and tested on out of time. The performance is decreasing slightly. Is there a way to supply posterior probabilities from May to aug data, using bayes implementation in SAS, to adjust the probabilities of Aug data?
03-14-2016 01:50 PM
You can specify a PRIOR dataset in your SCORE statement.
I'm not sure that's a valid way to develop a model.