Update metrics/mix_metrics.py

metrics/mix_metrics.py

@@ -0,0 +1,89 @@
+# metrics/mix_metrics.py
+
+import pandas as pd
+
+
+def calculate_count_mix(
+    df,
+    group_col,
+    id_col="account_id"
+):
+    """
+    Returns count mix across a characteristic.
+    """
+
+    result = (
+        df.groupby(group_col)[id_col]
+        .nunique()
+        .reset_index(name="accounts")
+    )
+
+    total_accounts = result["accounts"].sum()
+
+    result["mix_pct"] = (
+        result["accounts"] / total_accounts
+    ) * 100
+
+    result = result.sort_values(
+        by="accounts",
+        ascending=False
+    )
+
+    return result
+
+
+def calculate_limit_mix(
+    df,
+    group_col,
+    limit_col="credit_limit"
+):
+    """
+    Returns credit limit mix across a characteristic.
+    """
+
+    result = (
+        df.groupby(group_col)[limit_col]
+        .sum()
+        .reset_index(name="total_limit")
+    )
+
+    total_limit = result["total_limit"].sum()
+
+    result["mix_pct"] = (
+        result["total_limit"] / total_limit
+    ) * 100
+
+    result = result.sort_values(
+        by="total_limit",
+        ascending=False
+    )
+
+    return result
+
+
+def calculate_vintage_mix(
+    df,
+    vintage_col,
+    group_col,
+    id_col="account_id"
+):
+    """
+    Creates vintage-wise mix table.
+    """
+
+    result = (
+        df.groupby([vintage_col, group_col])[id_col]
+        .nunique()
+        .reset_index(name="accounts")
+    )
+
+    total_by_vintage = (
+        result.groupby(vintage_col)["accounts"]
+        .transform("sum")
+    )
+
+    result["mix_pct"] = (
+        result["accounts"] / total_by_vintage
+    ) * 100
+
+    return result