Section 3.3: Might be good to briefly mention thresholds for the Bayesian evidence when introducing it, considering that you then consider the Matern kernel to be “slightly disfavored”. I know this has a complicated answer, but you might simply say something on the lines of “Delta lnZ > 2 is a good threshold” and cite, e.g., Trotta 2008 (https://arxiv.org/abs/0803.4089). Something simple, that the reader can take your word for before looking at the plot.
dfm
commented
2 years ago
Section 3.3: Might be good to briefly mention thresholds for the Bayesian evidence when introducing it, considering that you then consider the Matern kernel to be “slightly disfavored”. I know this has a complicated answer, but you might simply say something on the lines of “Delta lnZ > 2 is a good threshold” and cite, e.g., Trotta 2008 (https://arxiv.org/abs/0803.4089). Something simple, that the reader can take your word for before looking at the plot.