Suppose a model is built to model the sales of 20 beer brands for thousands of stores across the U.S. The response variable is sales and the independent variables include marketing activity and macroeconomic factors. The model is built at the brand/store level. It would be difficult to estimate the thousands of brand/store intercepts individually and would be computational inefficient. Would mean centering the response variable and some of the macroeconomic variables allow one to "implicitly" calculate a brand/store intercept without explicitly including an intercept interaction in the model? Essentially would be accounting for differences in macroeconomic factors for each brand/store combination. You would have an estimate for a global intercept and than back calculate the brand/store level intercepts post estimation. Are there any papers that summarize this approach?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.