Day 74 (20:00): auto-format backend

Author: Test User

src/backend/api/chat.py

@@ -4974,7 +4974,12 @@ def send_message(
                 _q_pt = (
                     "classification"
                     if _q_algo.endswith("_classifier")
-                    or _q_algo in {"logistic_regression", "voting_classifier", "stacking_classifier"}
+                    or _q_algo
+                    in {
+                        "logistic_regression",
+                        "voting_classifier",
+                        "stacking_classifier",
+                    }
                     else "regression"
                 )
                 _q_rows = 0

src/backend/core/advisor.py

@@ -1363,7 +1363,9 @@ def compute_model_quality_score(
     """
     primary_metric, metric_name = _primary_metric(metrics, problem_type)
 
-    thresholds = _QUALITY_THRESHOLDS.get(problem_type, _QUALITY_THRESHOLDS["classification"])
+    thresholds = _QUALITY_THRESHOLDS.get(
+        problem_type, _QUALITY_THRESHOLDS["classification"]
+    )
 
     # Determine base quality band
     quality_label = "Needs Work"
@@ -1430,7 +1432,11 @@ def compute_model_quality_score(
         cv_pct_std = round(cv_std * 100, 1)
         reasoning.append(
             f"5-fold cross-validation: {cv_pct_mean}% ± {cv_pct_std}% — "
-            + ("consistent across splits." if is_stable else "results vary across splits.")
+            + (
+                "consistent across splits."
+                if is_stable
+                else "results vary across splits."
+            )
         )
 
     if stability_note:
@@ -1442,7 +1448,9 @@ def compute_model_quality_score(
     recommendation = _quality_recommendation(quality_label, problem_type, algorithm)
 
     # Final quality_score: interpolate within band
-    quality_score = _compute_quality_score(primary_metric, problem_type, base_score, quality_label)
+    quality_score = _compute_quality_score(
+        primary_metric, problem_type, base_score, quality_label
+    )
 
     return {
         "quality_label": quality_label,
@@ -1474,7 +1482,9 @@ def _compute_quality_score(
     quality_label: str,
 ) -> int:
     """Interpolate a 0–100 score within the quality band."""
-    thresholds = _QUALITY_THRESHOLDS.get(problem_type, _QUALITY_THRESHOLDS["classification"])
+    thresholds = _QUALITY_THRESHOLDS.get(
+        problem_type, _QUALITY_THRESHOLDS["classification"]
+    )
     # Find band boundaries
     band_min = 0.0
     band_max = 1.0
@@ -1513,7 +1523,15 @@ def _quality_recommendation(
     if quality_label == "Acceptable":
         is_nonlinear = any(
             kw in algorithm
-            for kw in ("forest", "gradient", "xgboost", "lgbm", "ensemble", "stacking", "voting")
+            for kw in (
+                "forest",
+                "gradient",
+                "xgboost",
+                "lgbm",
+                "ensemble",
+                "stacking",
+                "voting",
+            )
         )
         if not is_nonlinear:
             return (

src/backend/tests/test_model_quality_score.py

@@ -98,7 +98,9 @@ def test_cv_instability_downgrades_label():
     from core.advisor import compute_model_quality_score
 
     # R²=0.90 would normally be Excellent; with cv_std=0.15 → Good
-    r = compute_model_quality_score({"r2": 0.90, "cv_std": 0.15, "cv_mean": 0.85}, "regression")
+    r = compute_model_quality_score(
+        {"r2": 0.90, "cv_std": 0.15, "cv_mean": 0.85}, "regression"
+    )
     assert r["quality_label"] == "Good"
     assert r["is_stable"] is False
 
@@ -107,18 +109,24 @@ def test_cv_stability_no_downgrade_when_low():
     """Low cv_std (≤ 0.10) does not downgrade."""
     from core.advisor import compute_model_quality_score
 
-    r = compute_model_quality_score({"r2": 0.90, "cv_std": 0.05, "cv_mean": 0.88}, "regression")
+    r = compute_model_quality_score(
+        {"r2": 0.90, "cv_std": 0.05, "cv_mean": 0.88}, "regression"
+    )
     assert r["quality_label"] == "Excellent"
     assert r["is_stable"] is True
 
 
 def test_reasoning_bullets_present():
     from core.advisor import compute_model_quality_score
 
-    r = compute_model_quality_score({"r2": 0.73, "cv_mean": 0.71, "cv_std": 0.04}, "regression")
+    r = compute_model_quality_score(
+        {"r2": 0.73, "cv_mean": 0.71, "cv_std": 0.04}, "regression"
+    )
     assert len(r["reasoning"]) >= 2
     assert any("R²" in b or "73" in b for b in r["reasoning"])
-    assert any("cross-validation" in b.lower() or "cv" in b.lower() for b in r["reasoning"])
+    assert any(
+        "cross-validation" in b.lower() or "cv" in b.lower() for b in r["reasoning"]
+    )
 
 
 def test_recommendation_populated():
@@ -143,7 +151,10 @@ def test_nonlinear_recommendation_for_linear_algo():
         {"accuracy": 0.72}, "classification", algorithm="logistic_regression"
     )
     # Acceptable + linear algo → suggest nonlinear algorithm
-    assert "random" in r["recommendation"].lower() or "xgboost" in r["recommendation"].lower()
+    assert (
+        "random" in r["recommendation"].lower()
+        or "xgboost" in r["recommendation"].lower()
+    )
 
 
 def test_nonlinear_recommendation_skipped_for_ensemble():
@@ -167,7 +178,9 @@ def test_quality_score_within_range():
 def test_cv_fields_returned():
     from core.advisor import compute_model_quality_score
 
-    r = compute_model_quality_score({"r2": 0.80, "cv_mean": 0.78, "cv_std": 0.03}, "regression")
+    r = compute_model_quality_score(
+        {"r2": 0.80, "cv_mean": 0.78, "cv_std": 0.03}, "regression"
+    )
     assert r["cv_mean"] == pytest.approx(0.78)
     assert r["cv_std"] == pytest.approx(0.03)