Including more parameters to a given design commonly increases the R2 but can result in overfitting, i.e., managing the noise in the information as legitimate signal. When overfitting happens, the model will arrive at an incredibly large R2 from the information on which it was educated, but much decreased R2 https://josuegwkbm.blogdemls.com/32315881/5-tips-about-custom-fit-windows-you-can-use-today