fix: data_substitutions

Author: dshemetov

R/utils.R

@@ -155,19 +155,27 @@ get_exclusions <- function(
 }
 
 data_substitutions <- function(dataset, disease, forecast_generation_date) {
-  disease <- "flu"
+  # Get the substitutions from the table, matched by forecast generation date
   substitutions <- readr::read_csv(
     glue::glue("{disease}_data_substitutions.csv"),
     comment = "#",
     show_col_types = FALSE
   ) %>%
     filter(forecast_date == forecast_generation_date) %>%
     select(-forecast_date) %>%
-    rename(new_value = value)
-  dataset %>%
-    left_join(substitutions) %>%
+    rename(new_value = value) %>%
+    select(-time_value)
+  # Replace the most recent values in the appropriate keys with the substitutions
+  new_values <- dataset %>%
+    group_by(geo_value) %>%
+    slice_max(time_value) %>%
+    inner_join(substitutions) %>%
     mutate(value = ifelse(!is.na(new_value), new_value, value)) %>%
     select(-new_value)
+  # Remove keys from dataset that have been substituted
+  dataset %>%
+    anti_join(new_values, by = c("geo_value", "time_value")) %>%
+    bind_rows(new_values)
 }
 
 parse_prod_weights <- function(filename = here::here("covid_geo_exclusions.csv"),