coefficient standardization

bettertony

New Member
Hi All,

I am working on a 2nd order regression. The span of coefficient is large (say 1:10000). So the coefficients of 2nd order are very big and not in the same order of the 1st. I think it is needed to implement normalization (both coefficients and data). However, after I did the normalization (minus mean and divided by standard deviation), I found the coefficients of 2nd order and 1st order are almost the same (highly correlated). How can I deal with this? Am I wrong to do normalization?

Thanks.