Re: [R-br] percentage prediction error