We have some data. We have one model convertible into linear and non-linear form. I want to ask if we model the same data after conversion, results for both linear and non-linear forms for the same model will differ or not? In simple words will the R-square value for the linear and non-linear form differ or not? If it differs, I want to know how will it differ?
Dear Abhishek Ajmani,
If we normalize the non-Gaussian data by a bijective normalizing transformation, then the R-square values for the linear regression equation for the normalized data and the non-linear regression equation for the non-Gaussian data will not differ. If for normalization we use a non-bijective transformation, then the R-square values will be different. In this case, usually the R-square value for the linear regression equation is greater than the R-square value for the non-linear regression equation.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression