Re: [R] Using "line" (Tukey's robust line) on 3 observations?

Le 4 d�c. 2014 � 13:40, Tal Galili a �crit : > By accident I came across the following example: > > x <- 1:3 > y <- 1:3 > line(x, y) # returns: > > Call: > line(x, x) > > Coefficients: > [1] -2 2 > > > While when using 1:4, it will give the more reasonable 0,1 coefficients. > > I imagine

[R] Using "line" (Tukey's robust line) on 3 observations?

By accident I came across the following example: x <- 1:3 y <- 1:3 line(x, y) # returns: Call: line(x, x) Coefficients: [1] -2 2 While when using 1:4, it will give the more reasonable 0,1 coefficients. I imagine this is in the way it calculate the quantiles (i.e.: a "feature"). Can someone