Initial code for flagging rows for winsorisation

ONSdigital · Jun 20, 2024 · f2fa736 · f2fa736
1 parent 3280bec
commit f2fa736
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 0 deletions.
diff --git a/src/flag_for_winsorisation.py b/src/flag_for_winsorisation.py
diff --git a/tests/data/winsorisation/data_to_flag.csv → tests/data/winsorisation/flag_data.csv b/tests/data/winsorisation/data_to_flag.csv → tests/data/winsorisation/flag_data.csv
diff --git a/tests/test_flag_for_winsorisation.py b/tests/test_flag_for_winsorisation.py
@@ -0,0 +1,31 @@
+""" What we need
+
+dataframe
+a weight
+g weight
+
+a * g to create a new column
+
+then flag anything <=1 as not to be winsorised
+
+"""
+
+import pandas as pd
+
+
+def winsorisation_flag(df, a_weight, g_weight):
+
+    df["new_col"] = df.a_weight * df.g_weight
+
+    df["NW_AG_flag"] = df["new_col"].apply(lambda x: "NW_AG" if x <= 1 else "")
+
+    return df
+
+
+data = pd.read_csv(
+    "/home/cdsw/monthly-business-survey-results/tests/data/winsorisation/flag_data.csv"
+)
+
+print(data)
+
+test = winsorisation_flag(data, data.a_weight, data.g_weight)