|
have a great doubt whether I am doing correctly. My goal is to perform linear regression, and I have Y as a response variable and X_1 and X_2 as explanatory variables. The model should be nonlinear, so I applied a simple nonlinear transformation Y = b_0 + b_1 X_1 + b_2 X_2 + b_3 X_1^2 + b_4 X_2^2 + b_5 X_1X_2 But, since the range of both X_1,X_2 is about [600,1500], the quadratic terms have a very large range, so I decided to normalize. My question is: should I normalize after applied the nonlinear transformation, or should I normalize the original variables X_1 and X_2 and then apply the nonlinear transformation on the normalized variables? Thanks |
