Update app.py

app.py

@@ -382,6 +382,127 @@ def generate_portfolio_summary():
     except Exception as e:
         return f"Error generating summary: {str(e)}"
 
+
+# ---------------------------------------------------
+# PORTFOLIO OVERVIEW (Calendar Snapshot)
+# ---------------------------------------------------
+
+
+def _detect_date_column(df):
+    candidates = [
+        "observation_date",
+        "observation_month",
+        "obs_date",
+        "date",
+        "calendar_month",
+        "month",
+        "report_date"
+    ]
+    for c in candidates:
+        if c in df.columns:
+            return c
+    return None
+
+
+def get_calendar_months():
+    date_col = _detect_date_column(master_df)
+    if date_col is None:
+        return []
+    ser = pd.to_datetime(master_df[date_col], errors="coerce")
+    months = ser.dt.to_period("M").astype(str).dropna().unique().tolist()
+    months.sort()
+    return months
+
+
+def _filter_master_by_month(as_of_month):
+    # as_of_month expected like "YYYY-MM"
+    date_col = _detect_date_column(master_df)
+    if date_col is None or not as_of_month:
+        return master_df.copy()
+    ser = pd.to_datetime(master_df[date_col], errors="coerce").dt.to_period("M").astype(str)
+    return master_df[ser == as_of_month].copy()
+
+
+def generate_portfolio_overview(as_of_month, segment):
+    """
+    Returns a small DataFrame with key portfolio snapshot metrics for the selected calendar month and segment.
+    Metrics: Total Accounts, Open Accounts (balance>0), Bad Accounts (dpd>=30), Overall NCL Rate (dollar %), Average FICO.
+    """
+    df = _filter_master_by_month(as_of_month)
+
+    # If segment filtering required in future, `segment` param can be used to filter further
+
+    total_accounts = df["account_id"].nunique() if "account_id" in df.columns else 0
+
+    open_accounts = (
+        df.loc[df["balance"] > 0, "account_id"].nunique()
+        if "balance" in df.columns
+        else total_accounts
+    )
+
+    bad_accounts = (
+        df.loc[df["dpd"].fillna(0) >= 30, "account_id"].nunique()
+        if "dpd" in df.columns
+        else 0
+    )
+
+    # overall NCL rate (dollar-based) fallback logic
+    ncl_cols = [c for c in df.columns if "ncl" in c.lower()]
+    overall_ncl_rate = None
+    if len(ncl_cols) > 0 and "balance" in df.columns:
+        ncl_sum = df[ncl_cols[0]].sum(skipna=True)
+        bal_sum = df["balance"].sum(skipna=True)
+        overall_ncl_rate = (ncl_sum / bal_sum * 100) if bal_sum > 0 else None
+    else:
+        # fallback: use bad balance / total balance as proxy
+        if "balance" in df.columns and "dpd" in df.columns:
+            bad_bal = df.loc[df["dpd"].fillna(0) >= 30, "balance"].sum()
+            bal_sum = df["balance"].sum()
+            overall_ncl_rate = (bad_bal / bal_sum * 100) if bal_sum > 0 else None
+
+    if overall_ncl_rate is None:
+        overall_ncl_rate = float("nan")
+    else:
+        overall_ncl_rate = round(overall_ncl_rate, 2)
+
+    # average fico
+    avg_fico = None
+    if "fico_score" in df.columns:
+        avg_fico = round(df["fico_score"].dropna().mean(), 1)
+    elif "fico_band" in df.columns:
+        # attempt to map band midpoints like "600-650" -> 625
+        def band_mid(b):
+            try:
+                parts = b.split("-")
+                return (int(parts[0]) + int(parts[1])) / 2
+            except Exception:
+                return None
+        mid_vals = df["fico_band"].dropna().apply(band_mid).dropna()
+        avg_fico = round(mid_vals.mean(), 1) if not mid_vals.empty else float("nan")
+    else:
+        avg_fico = float("nan")
+
+    overview = pd.DataFrame({
+        "Metric": [
+            "As Of Month",
+            "Total Accounts",
+            "Open Accounts",
+            "Bad Accounts (dpd>=30)",
+            "Overall NCL Rate (%)",
+            "Average FICO"
+        ],
+        "Value": [
+            as_of_month if as_of_month else "All",
+            int(total_accounts),
+            int(open_accounts),
+            int(bad_accounts),
+            overall_ncl_rate,
+            avg_fico
+        ]
+    })
+
+    return overview
+
 # ---------------------------------------------------
 # DYNAMIC DROPDOWNS
 # ---------------------------------------------------
@@ -820,4 +941,40 @@ with gr.Blocks() as app:
                 outputs=summary_table
             )
 
+        # =================================================
+        # TAB 4: PORTFOLIO OVERVIEW (Calendar Snapshot)
+        # =================================================
+
+        with gr.TabItem("📅 Portfolio Overview"):
+
+            gr.Markdown("## Portfolio Snapshot by Calendar Month")
+
+            with gr.Row():
+                calendar_month_dropdown = gr.Dropdown(
+                    choices=get_calendar_months(),
+                    value=(get_calendar_months()[-1] if len(get_calendar_months()) > 0 else None),
+                    label="Calendar Month (YYYY-MM)"
+                )
+
+                overview_segment_dropdown = gr.Dropdown(
+                    choices=[
+                        "fico_band",
+                        "sourcing_channel",
+                        "city_tier",
+                        "occupation_type"
+                    ],
+                    value="fico_band",
+                    label="Segment (for drill)"
+                )
+
+                gen_overview_btn = gr.Button("Generate Snapshot", variant="primary")
+
+            overview_table = gr.Dataframe(label="Portfolio Overview")
+
+            gen_overview_btn.click(
+                fn=generate_portfolio_overview,
+                inputs=[calendar_month_dropdown, overview_segment_dropdown],
+                outputs=overview_table
+            )
+
 app.launch()