XX' matrix not always invertible

[Alain38]

@PaigeMiller@Rick_SASThank you very much for help, I have now more elements to decide.

Don't compute the regression for this data, instead print an error message of some sort.

Knowing that I have a loop, how can I say: "If the X'X matrix is singular, then do not abort but write for example #N/A" so that when I look at the output table, I can see that for, let say t = 350 and t = 580, the matrix was not singular and these particular data points will not be used ?

[Rick_SAS]

if abs(det(A)) < 1e-8 then 

   /* handle singular case */

else

   /* handle nonsingular case */

[Rick_SAS]

For more about generalized inverses and solving the rank-deficient regression problem, see the articles

"Generalized inverses for matrices" and

"Singular parameterizations, generalized inverses, and regression estimates"