@caufield wrote:
Hi - I am re-building a relatively simple credit risk style logistic regression model, currently the only explanatory variable is a score from another model (predicting something similar but not identical), I am then checking misalignment of any other key variables and adding any in as required.
My question is; Is there anything fundamentally incorrect about using a “score” from another model, within a new model? And then subsequently performing a misalignment analysis on some more variables to potentially include them?
Posts: 1
Participants: 1