search

fhui28 / CBFM

Spatio-temporal joint species distribution modeling using community-level basis functions
13 stars 2 forks source link

Adapting k.check for spatio-temporal basis functions? #10

Closed fhui28 closed 5 months ago

fhui28 commented 3 years ago

Currently advice in main help file is provided to plot residuals as a smooth function of the spatial/temporal coordinates (with large k) using a gam, say, and see whether there is substantial residual pattern here. This is adapted from the help file of choose.k.

However, perhaps the simple reshuffling residual diagnostic test could also be implemented for B_space/B_time/B_spacetime?

fhui28 commented 5 months ago

Perhaps this is not a good idea in the long term, and it is better instead to leave it more grey and more trial-and-error.