Dear SPMers,

 

I am learning non-parametric methods in SnPM. I found that SnPM use [ regressors - mean(regressors) ] as labels to do regression instead of just use regressors as labels. Is there any reason to do that?

 

Thanks,

Xue