GP with Multiple input and single output

Hi Fellows,

I am interested in learning GP with multiple input, however i could not find any relevant documents or examples thay was done by stan to help me the concepts behind. I am wondering if anyone across with such problem to help me with learning materials

1 Like

There is a chapter in the stan manual 10.1 Gaussian process regression | Stan User’s Guide

The section on the manual doesn’t include a full worked example with multiply input variables: 10.2 Simulating from a Gaussian process | Stan User’s Guide. It includes a description of how to modify the input data structure but then concludes with:

In the remainder of the chapter, univariate models will be used for simplicity, but any of the models could be changed to multivariate in the same way as the simple sampling model. The only extra computational overhead from a multivariate model is in the distance calculation.

But this is not an obvious extension if you are new to GPs. Indeed, even beyond Stan it’s very hard to find worked examples of this situation (I was recently looking for this myself and struggled to find worked examples beyond univariate in Python, R or Julia code). Anyhow, I guess what I’m saying is the manual might benefit from including such an example in full.