Covariance of parameter estimates in nonlinear mixed effects fit

2 views (last 30 days)
Sanjay Manohar
Sanjay Manohar on 30 Jul 2015
Edited: Sanjay Manohar on 30 Jul 2015
I'm using nlmefit, and am trying to test the hypotheses using linhyptest. For that I need the covariance of the parameter estimates. The documentation suggests that the "stats" output structure should have a field called "covb". My output structure does not contain this!
[beta,psi,stats,re_subj] = nlmefit( X , Y, [],
@(b,X)X*b', [0,0,0,0], 'REParam',[1] );
stats =
dfe: 4844
logl: -1.1235e+04
mse: 5.9078
rmse: 2.4274
errorparam: 2.4306
aic: 2.2482e+04
bic: 2.2488e+04
sebeta: [0.4017 0.0919 0.0426 0.0426]
How can I get the required matrix? (I was expecting a 4x4, with 'sebeta.^2' as diagonals maybe?)

Answers (0)

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!