search

RuiZhang2016 / Variational-Inference-for-SGP-Modulated-Hawkes-Process

This repo contains a demo for the AAAI-20 publication "Variational Inference for Sparse Gaussian Process Modulated Hawkes Process"
MIT License
5 stars 2 forks source link

Get predictive log likelihood for custom data #2

Open manishadubey91 opened 2 years ago

manishadubey91 commented 2 years ago

Could you please tell that how can I get predictive scores for a custom data?

RuiZhang2016 commented 2 years ago

Hi,

You could look into the demo jupyter notebook where the GP model is drawn. Use the predictive mean to compute the likelihood on the test data as the usual case or use ''TELBO'' in vbhp.log_marginal_likehood(). It is an approximation to the marginal likelihood of the Bayesian model so figure out whether it is what you want before using it :).