0

Does anyone know if it is possible to fit a "logistic Gaussian process regression model" with INLA?

The model can be found in the below link:

https://mc-stan.org/docs/2_19/stan-users-guide/fit-gp-section.html

It simply works by replacing predictors linear effect with a GP to make Logistic regression non-linear. It's computationally hard to fit it using MCMC so I am wondering if it's possible to fit it using INLA.

Vinícius Félix
  • 8,448
  • 6
  • 16
  • 32
Amin Shn
  • 532
  • 2
  • 11
  • what K() (autocorrelation function) do you want to use? – Ben Bolker Sep 19 '21 at 23:49
  • @BenBolker I am really not sure about that maybe "exponentiated quadratic kernel", I just want to replace XB (linear effect of multiple predictors) in logit(p) = X.B with a GP to get a non-linear boundary separating classes in the feature space and hopefully get higher classification accuracy. Any suggestion? – Amin Shn Sep 20 '21 at 07:42

0 Answers0