Delimit big counts with commas

analysis/report/report_utils.py

@@ -324,8 +324,8 @@ def format_season_table(df, column_to_plot, category):
     rr_cis_str = ci_to_str(rr_cis)
     rr_cis_str.name = ("2023 v 2018", "Rate Ratio (95% CI)")
 
-    # Create table with count, rate (95% CI)
-    table["Count"] = table["Count"].applymap(lambda x: f"{x:.0f}")
+    # Create table with (comma delimited) count, rate (95% CI)
+    table["Count"] = table["Count"].applymap(lambda x: f"{x:,.0f}")
     count_cis = (table[["Count", "Rate (95% CI)"]]).swaplevel(axis=1)
     cols = count_cis.columns.get_level_values(0).unique()
     reordered = count_cis.reindex(columns=cols, level="gas_year")

analysis/report/table1.py

@@ -51,8 +51,11 @@ def series_to_bool(series):
 
 
 def transform_percentage(x):
+    """
+    Return (comma delimited) x and column percentage
+    """
     transformed = (
-        x.map("{:.0f}".format)
+        x.map("{:,.0f}".format)
         + " ("
         + (((x / x.sum()) * 100).round(1)).astype(str)
         + ")"

analysis/report/tables.py

@@ -7,17 +7,7 @@
     subset_table,
 )
 import pandas
-
-
-def round_or_skip(x):
-    """
-    Try to return a value
-    If it fails, return the original value
-    """
-    try:
-        return round(float(x))
-    except Exception:
-        return x
+import numpy
 
 
 def count_table(measure_table):
@@ -96,10 +86,10 @@ def main():
     table = count_table(subset)
     table.index.name = table.index.name.title()
     table.index = table.index.strftime("%d-%m-%y")
-    # NOTE: applymap is by value (so very slow), but we have a mix of strings
-    # and numbers (REDACTED) so pandas.to_numeric will not work
-    # and the tables are small
-    table = table.applymap(round_or_skip)
+    # NOTE: applymap is by value (so very slow), but the tables are small
+    table = table.applymap(
+        lambda x: f"{x:,.0f}" if ~numpy.isnan(x) else "[REDACTED]"
+    )
     table.to_csv(output_dir / output_name)