Covariance estimate using fitrgp

I understand that it's possible to obtain a 95% confidence interval using [ypred1,~,yint1] = predict(gprMdl1,x); but is it possible to directly obtain an estimate of covariance/uncertainty statistics over the evaluation points?

Answer 1

[Ayush Anand]

Hi George,

I understand you want to know if there is a way of directly obtaining an estimate of covariance over the evaluation points. Unfortunately, there is no function in MATLAB which computes this directly; you can use the predicted mean and the actual value at the evaluation points to compute the covariance/uncertainty according to the standard formula for computing variance.

I hope this helps!