add data former members of czechoslovakia and yugoslavia

owid · May 13, 2024 · f706500 · f706500
1 parent a8fdc9f
commit f706500
Show file tree

Hide file tree

Showing 3 changed files with 48 additions and 3 deletions.
diff --git a/etl/steps/data/garden/democracy/2024-03-07/lexical_index.countries_impute.yml b/etl/steps/data/garden/democracy/2024-03-07/lexical_index.countries_impute.yml
@@ -67,3 +67,43 @@
   country_impute: Russia
   year_min: 1813
   year_max: 1921
+
+# Czechoslovakia
+- country:
+    - Czechia
+    - Slovakia
+  country_impute: Czechoslovakia
+  year_min: 1918
+  year_max: 1992
+
+# Yugoslavia
+- country:
+    - Croatia
+    - Bosnia and Herzegovina
+    - Slovenia
+    - North Macedonia
+  country_impute: Yugoslavia
+  year_min: 1918
+  year_max: 1989
+- country:
+    - Serbia
+    - Montenegro
+    - Kosovo
+  country_impute: Yugoslavia
+  year_min: 1918
+  year_max: 1991
+
+# Serbia and Montenegro
+- country:
+    - Serbia
+    - Montenegro
+    - Kosovo
+  country_impute: Serbia and Montenegro
+  year_min: 1992
+  year_max: 2005
+
+# Kosovo
+- country: Kosovo
+  country_impute: Serbia
+  year_min: 2006
+  year_max: 2007
diff --git a/etl/steps/data/garden/democracy/2024-03-07/lexical_index.meta.yml b/etl/steps/data/garden/democracy/2024-03-07/lexical_index.meta.yml
@@ -16,6 +16,10 @@ tables:
     common:
       description_processing: |-
         Data for former USSR member states has been obtained by imputing the values of the USSR. This includes: Russia, Ukraine, Belarus, Latvia, Lithuania, Estonia, Armenia, Azerbaijan, Georgia, Kazakhstan, Kyrgyzstan, Tajikistan, Turkmenistan, Uzbekistan, and Moldova.
+
+        Data for former Czechoslovakia member states has been obtained by imputing the values of Czechoslovakia. This includes: Czech Republic and Slovakia.
+
+        Data for former Yugoslavia member states has been obtained by imputing the values of Yugoslavia. This includes: Slovenia, Croatia, Bosnia and Herzegovina, Serbia, Montenegro, Kosovo, and North Macedonia.
     variables:
       exelec_lied:
         title: Elections for chief executive

diff --git a/etl/steps/data/garden/democracy/2024-03-07/lexical_index.py b/etl/steps/data/garden/democracy/2024-03-07/lexical_index.py
@@ -123,14 +123,15 @@ def run(dest_dir: str) -> None:
     ) == {"Germany"}, "Other versions of Germany!"
 
     # Impute values
-    tb = add_imputes(tb=tb, path=PATH_IMPUTE, col_flag_imputed="values_imputed")
+    col_flag_imputed = "values_imputed"
+    tb = add_imputes(tb=tb, path=PATH_IMPUTE, col_flag_imputed=col_flag_imputed)
 
     # Get region data
-    tb_regions = tb.loc[~tb["values_imputed"]].drop(columns=["values_imputed"]).copy()
+    tb_regions = tb.loc[~tb[col_flag_imputed]].drop(columns=[col_flag_imputed]).copy()
     tb_regions = get_region_aggregates(tb_regions, ds_regions, ds_population)
 
     # Drop is imputed flag
-    tb = tb.drop(columns=["values_imputed"])
+    tb = tb.drop(columns=[col_flag_imputed])
 
     # Format
     tb = tb.format(["country", "year"])