JuliaGaussianProcesses / GPLikelihoods.jl

Provides likelihood functions for Gaussian Processes.
https://juliagaussianprocesses.github.io/GPLikelihoods.jl/
MIT License
43 stars 5 forks source link

Question: Scaled Poisson expectation #67

Open filipepfarias opened 2 years ago

filipepfarias commented 2 years ago

I'm trying to implement the GP with a Poisson likelihood whose parameter may vary with the observations[^1]: $$ y_i \sim Poisson(e_i exp(f_i))$$ where f_i is the GP. Is this possible?

[^1]: Approximate inference for disease mapping with sparse Gaussian processes, Jarno Vanhatalo, Ville Pietiläinen, Aki Vehtari (Eq. 1)

theogf commented 2 years ago

Hi,

It is not possible right now and having external parameters is something we haven't considered so far but definitely should!