All workflow information to reproduce key results of the paper is contained in reproduce.Rmd. The steps for running the workflow are:
Step 1: Install all required R packages in dependencies.txt.
Step 2 (Optional): Compile the source code for the BRISC library by running the following command in your terminal:
R CMD SHLIB BRISC/BRISC.cpp BRISC/BRISC_Predict.cpp BRISC/util.cpp BRISC/init.c BRISC/lbfgs.c
This step is optional and only required if you want to re-run SPGP model fitting and prediction.
Step 3 (Optional): Modify the options in Line 19-23 of reproduce.Rmd accordingly. By default, key figures and tables are reproduced from pre-computed model outputs in data/. To re-generate model outputs, ma
5B4B
ke the following change(s):
- To re-generate simulation input data, set
gen_sim_input = TRUE. - To re-run SPGP model in simulation, set
fit_SPGP_sim = TRUE. - To re-run TGP model in simulation, set
fit_TGP_sim = TRUE. - To re-run SPGP model in precipitation data application, set
fit_SPGP_app = TRUE. - To re-run TGP model in precipitation data application, set
fit_TGP_app = TRUE.
Please note that re-running these models can take several hours.
Step 4: Run the workflow in reproduce.Rmd by compiling the R Markdown file. For example, this can be done by running the following command in your terminal:
Rscript -e "rmarkdown::render('reproduce.Rmd')"
Reproducible results can be found in the generated reproduce.html file.