what is exactly yhat in output of regstat ?

Hello

I use regstats to fit a nonlinear trend. I just check as an example the firt using a linear fit with regstats. X is my time series (length = 2000).. I used s=regstats([1:2000],X,'linear').

Then I am surprised that s.yhat is ... non linear while the fitted value (a straight line in my case) is retrieven accurately using fit=s.beta(1)*[1:2000]+s.beta(2);

So what is exactly s.yhat ?

Thanks in advance

Vincent

Answer 1

[Tom Lane]

Looks to me like you reversed the two inputs to regstats. Unfortunately this function takes (y,x) rather then (x,y):

>> help regstats
 regstats Regression diagnostics for linear models.
    regstats(RESPONSES,DATA,MODEL) regresses measurements in the vector