I've added https://sourceforge.net/p/hsdm/code/ci/dev/tree/examples/hSDM.nc_examples.R this example script to illustrate this the existing functions for two simple models. The challenge will be to make the hSDM.nc function work will all types of models and options. But I think using netcdf and embedding metadata (species, model details, etc.) will be a really useful addition for big (multi-model, multi-species) projects.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
--- old+++ new@@ -1,2 +1,2 @@-I'm adding functions to write hSDM output to netcdf format to facilitate multi-model and multi-species model comparisons and evaluation. +Adding functions to write hSDM output to netcdf format to facilitate multi-model and multi-species model comparisons and evaluation.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
The current version hSDM.nc works, but creates a complete copy of the input data in each run of each model of output - wasteful. So perhaps better to to write a 'modelinput' file with the input (covariates and occurrence) data instead of embedding it all in the each output file (which is wasteful for multi-model comparisons in large regions).
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I've added https://sourceforge.net/p/hsdm/code/ci/dev/tree/examples/hSDM.nc_examples.R this example script to illustrate this the existing functions for two simple models. The challenge will be to make the hSDM.nc function work will all types of models and options. But I think using netcdf and embedding metadata (species, model details, etc.) will be a really useful addition for big (multi-model, multi-species) projects.
Diff:
The current version hSDM.nc works, but creates a complete copy of the input data in each run of each model of output - wasteful. So perhaps better to to write a 'modelinput' file with the input (covariates and occurrence) data instead of embedding it all in the each output file (which is wasteful for multi-model comparisons in large regions).