search for: sefit

...abs(x1 - x0) # distance from focal x h <- sort(dist1)[.4*length(x1)] # bandwidth for span: bandwidth = span*length(x1) inwindow <- dist1 <= h # observations within window d <- dist1[dist1 <= h] z <- d/h w1 <- (1-abs(z)^3)^3 w2 <- sum(w1^2) s*sqrt(w2) # calculated sefit > s*sqrt(w2) [1] 8.052887 this is quite high, and r gives me > predict(lo1, se=T)$se.fit[1] 1 0.6055714 although I did calculate s correctly > lo1$s [1] 1.484307 > s [1] 1.484307 So clearly something is wrong with the weight part of my equation. I'm not sure whether...