Closed pedroliman closed 2 years ago
LIke in our COVID-19 work, runs should be ordered to allow for comparisons across scenarios as soon as possible.
This is how we did it in the covid-19 work:
sow_experiments = unique(experiment_model_rdm$future_experimental_design %>% select(RunID, LHSExperimentID)) # Randomizing the order of the sow_experiments: sow_experiments = sow_experiments[sample(nrow(sow_experiments), replace = F),] # adding the grid experiment ids: GridExperimentID = unique(experiment_model_rdm$future_experimental_design$GridExperimentID) experiments_order = tidyr::expand_grid(sow_experiments, GridExperimentID) future_experimental_design = experiments_order %>% dplyr::left_join(experiment_model_rdm$future_experimental_design, by = c("RunID", "LHSExperimentID", "GridExperimentID")) %>% mutate(ExperimentID = row_number()) experiment_model_rdm$future_experimental_design = future_experimental_design
This is no longer required given how the experimental design is created. The experimental design is already created in an order that will allow early comparisons between policies.
LIke in our COVID-19 work, runs should be ordered to allow for comparisons across scenarios as soon as possible.
This is how we did it in the covid-19 work: