> > x wrote: > > > x2 NA 1.797e+14 NA NA > > > x2' NA 6.475e+14 NA > NA > > > > > > Residual standard error: 82.44 on 95 degrees of > > freedom > > > Adjusted R-Squared: 0.9992 > > > Error in if (coef[i] > 0 & (i > 2 | > coef[1] > > != 0 | Intc != 0)) "+" else NULL : missing > value > > where TRUE/FALSE needed > > > It just appears that you have perfect prediction, so > you > > have quite an unusual dataset to be doing inference > on. > > > > Frank > > OK, I tried different test datasets. > > 1) y1 <- rnorm(1000); x1 <- runif(1000); x2 <- > runif(1000); > does NOT show any error. > > 2) y1 = 10*x1*e1 + 10*x2*e1 + 100*e1 where x1=1..100, > x2=100..1, e1=Gaussian noise > gives me error like before. > > 3) y1 = x1^3*e1 + x2^3*e1 + 100*e1^2 also gives me the same > error. > > Now, why should there be perfect prediction in cases 2 > & 3?
I tried another test - I replicated x1 to be x2 as well. While y1~rcs(x1,3) works y1~( rcs(x1,3) + rcs(x2,3) ) does not (error message shown above). Thanks again, sp ______________________________________________ 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.
