I need to perform a semi-log regression (or log-level regression) for sales data.

For that, i choosed a theoretical model which matches with my regression goal. It looks like the following:

Y = exp(X1)+exp(X2)+exp(X3).. and so on


Y=Absolute Number of sales for brand i

X1 = Baseprice for brand i

X2...XN = Baseprice for every other brand

I know that the model should lead to coefficients that can be interpreted as elasticities (I need to get Own- and Cross-Price-Elasticities). But my Coefficients are much greater than 1 and the R-Squared Value indicates an extremely bad fit for some brands.

Maybe i am completely wrong with the interpretation? Maybe the fool is in the data?

Please excuse the wide ranged question, but i really dont know where to start fitting the model, so i am very thankful for any advice



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.