Added dropped sites.

6-statistics-fusion-order.py

@@ -6,7 +6,8 @@ Inputs (``data/``, ``{year}`` = ``--evaluation-year``):
 
 Outputs (``data/statistics_fusion_order/``):
 
-- ``{year}.json`` — paired Wilcoxon + t-test summary for NSE, RMSE, nRMSE, r
+- ``{year}.json`` — paired Wilcoxon + t-test summary for NSE, RMSE, nRMSE, r;
+  includes ``dropped_sites`` (union) and per-metric ``dropped_sites`` lists
 
 CLI:
 
@@ -50,48 +51,48 @@ def _r4(v: float | None) -> float | None:
     return round(v, 4) if v is not None else None
 
 
-def _load_site_metrics(year: int) -> list[dict[str, Any]]:
-    """Return parsed ``metrics.json`` payloads for every site under ``{year}``."""
+def _load_site_metrics(year: int) -> list[tuple[str, dict[str, Any]]]:
+    """Return ``(sitename, metrics.json payload)`` for every site under ``{year}``."""
     metrics_dir = DATA_DIR / "metrics" / str(year)
     if not metrics_dir.is_dir():
         return []
 
-    payloads: list[dict[str, Any]] = []
+    payloads: list[tuple[str, dict[str, Any]]] = []
     for site_dir in sorted(metrics_dir.iterdir()):
         if not site_dir.is_dir():
             continue
         path = site_dir / "metrics.json"
         if not path.is_file():
             continue
-        payloads.append(json.loads(path.read_text()))
+        payloads.append((site_dir.name, json.loads(path.read_text())))
     return payloads
 
 
 def collect_pairs(
-    site_metrics: list[dict[str, Any]], metric: str
-) -> tuple[list[float], list[float], int]:
-    """Return paired BtI / ItB values for ``metric`` and count of dropped sites."""
+    site_metrics: list[tuple[str, dict[str, Any]]], metric: str
+) -> tuple[list[float], list[float], list[str]]:
+    """Return paired BtI / ItB values for ``metric`` and dropped site names."""
     bti_vals: list[float] = []
     itb_vals: list[float] = []
-    n_dropped = 0
+    dropped_sites: list[str] = []
 
-    for payload in site_metrics:
+    for site, payload in site_metrics:
         bti = payload.get("bti")
         itb = payload.get("itb")
         if not isinstance(bti, dict) or not isinstance(itb, dict):
-            n_dropped += 1
+            dropped_sites.append(site)
             continue
 
         bti_v = bti.get(metric)
         itb_v = itb.get(metric)
         if bti_v is None or itb_v is None:
-            n_dropped += 1
+            dropped_sites.append(site)
             continue
 
         bti_vals.append(float(bti_v))
         itb_vals.append(float(itb_v))
 
-    return bti_vals, itb_vals, n_dropped
+    return bti_vals, itb_vals, dropped_sites
 
 
 def _better_order(
@@ -226,16 +227,20 @@ def main() -> None:
         )
 
     metrics_out: dict[str, Any] = {}
+    all_dropped: set[str] = set()
     for metric in METRICS:
-        bti_vals, itb_vals, n_dropped = collect_pairs(site_metrics, metric)
+        bti_vals, itb_vals, dropped_sites = collect_pairs(site_metrics, metric)
         summary = paired_test(bti_vals, itb_vals, metric, alpha)
-        summary["n_dropped"] = n_dropped
+        summary["n_dropped"] = len(dropped_sites)
+        summary["dropped_sites"] = dropped_sites
+        all_dropped.update(dropped_sites)
         metrics_out[metric] = summary
 
     payload = {
         "year": year,
         "alpha": alpha,
         "n_sites_total": n_sites_total,
+        "dropped_sites": sorted(all_dropped),
         "metrics": metrics_out,
     }