Fill in using latest data first

R/adapt_source.R

@@ -103,30 +103,30 @@ adapt_source <- function(gdp, source, with_regions, replace_NAs, require_year_co
       dplyr::group_by(.data$iso3c) %>%
       tidyr::fill(c("MER (LCU per US$)",
                     "PPP conversion factor, GDP (LCU per international $)"),
-                  .direction = "downup") %>%
+                  .direction = "updown") %>%
       dplyr::ungroup() %>%
       # For the deflator, we need to multiply the bordering values with the actual USA growth
       dplyr::left_join(USA_def_growth, by = dplyr::join_by("year")) %>%
-      # Forward
+      # Backward
+      dplyr::arrange(-.data$year) %>%
       dplyr::mutate(
         `GDP deflator` = purrr::accumulate(
           dplyr::row_number(),
-          ~ dplyr::coalesce(.data$`GDP deflator`[.y], .x * .data$gd[.y]),
+          ~ dplyr::coalesce(.data$`GDP deflator`[.y], .x / .data$gd[.y]),
           .init = NA
         )[-1],
         .by = c("iso3c")
       ) %>%
-      # Backward
-      dplyr::arrange(-.data$year) %>%
+      dplyr::arrange(.data$iso3c, .data$year) %>%
+      # Forward
       dplyr::mutate(
         `GDP deflator` = purrr::accumulate(
           dplyr::row_number(),
-          ~ dplyr::coalesce(.data$`GDP deflator`[.y], .x / .data$gd[.y]),
+          ~ dplyr::coalesce(.data$`GDP deflator`[.y], .x * .data$gd[.y]),
           .init = NA
         )[-1],
         .by = c("iso3c")
       ) %>%
-      dplyr::arrange(.data$iso3c, .data$year) %>%
       dplyr::select(-"gd")
 
     # If there is no PPP data whatsoever for the country, use MERs