Add model selector for rank history

src/utils.py

@@ -806,6 +806,7 @@ BASELINE_RANK_HISTORY_COLUMNS = [
 _RANK_COLS = [c for c in BASELINE_RANK_HISTORY_COLUMNS if c.endswith("_Rank")]
 _DAILY_DISPLAY_COLS = ["date", "model"] + _RANK_COLS
 _WEEKLY_DISPLAY_COLS = ["week", "model"] + _RANK_COLS
+_MONTHLY_DISPLAY_COLS = ["month", "model"] + _RANK_COLS
 
 
 def _read_rank_history(path: Path, date_col: str) -> pd.DataFrame:
@@ -843,6 +844,47 @@ def load_baseline_rank_history_weekly(root_dir: str = "results") -> pd.DataFrame
     return _read_rank_history(Path(root_dir) / "baseline_rank_history_weekly.csv", "week")
 
 
+def load_baseline_rank_history_monthly(root_dir: str = "results") -> pd.DataFrame:
+    """Load monthly tracked-model rank history aggregated from daily history."""
+    daily_path = Path(root_dir) / "baseline_rank_history_daily.csv"
+    if not daily_path.exists():
+        daily_path = Path(root_dir) / "baseline_rank_history.csv"
+    if not daily_path.exists():
+        return pd.DataFrame(columns=_MONTHLY_DISPLAY_COLS)
+    try:
+        df = pd.read_csv(daily_path)
+    except Exception:
+        return pd.DataFrame(columns=_MONTHLY_DISPLAY_COLS)
+    if "date" not in df.columns:
+        return pd.DataFrame(columns=_MONTHLY_DISPLAY_COLS)
+
+    df["date"] = pd.to_datetime(df["date"], errors="coerce")
+    df = df.dropna(subset=["date"])
+    if df.empty:
+        return pd.DataFrame(columns=_MONTHLY_DISPLAY_COLS)
+    if "model" not in df.columns:
+        df["model"] = ""
+    df["month"] = df["date"].dt.strftime("%Y-%m")
+    for col in _RANK_COLS:
+        if col in df.columns:
+            df[col] = pd.to_numeric(df[col], errors="coerce")
+
+    present_rank_cols = [col for col in _RANK_COLS if col in df.columns]
+    if present_rank_cols:
+        monthly = (
+            df.groupby(["month", "model"], dropna=False)[present_rank_cols]
+            .mean(numeric_only=True)
+            .round(3)
+            .reset_index()
+        )
+    else:
+        monthly = df.loc[:, ["month", "model"]].drop_duplicates()
+    for col in present_rank_cols:
+        monthly[col] = monthly[col].map(format_rank_number)
+    present = [c for c in _MONTHLY_DISPLAY_COLS if c in monthly.columns]
+    return monthly[present].sort_values(["month", "model"]).reset_index(drop=True)
+
+
 # ---------------------------------------------------------------------------
 # Forecast snapshot utilities
 # ---------------------------------------------------------------------------