Extract linear model parameters from fitted linear gp
goldingn opened this issue · comments
issues:
- how to extract linear kernel from compositional kernel?
- write a function to spot additive bits and look at sub-kernels?
- then get all the ANOVA components for models, and write anova function for post-hoc analysis?
similarly, extract variance component and predicted levels from iid and intercept kernels - use kernel-specific projection for these (or set other data to be the same)