1) In my very humble opinion R^2 can't be negative, at least for data for which it sound to use linear model. Or the data would have to be utterly wrong to fit them with linear model.
2) I don't want to fit data with linear model of zero intercept. 3) I dont know if I understand correctly. Im 100% sure the model for my data should have zero intercept. The only coordinate which Im 100% sure is correct. If I had measured quality Y of a same sample X0 number of times I would get E(Y(X0))=0. Basically what I need to compute is R^2=1- (sum(residue^2))/(sum(Y[i])) both for model with and without intercept. I don't want to consider null model (model with zero intercept) at all. I don't know why to use y* in R^2 = 1 - Sum(R[i]^2) / Sum((y[i]- y*)^2) -- View this message in context: http://r.789695.n4.nabble.com/Strange-R-squared-possible-error-tp3382818p3383277.html Sent from the R help mailing list archive at Nabble.com. ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
