Create scorer.py

scorer.py

@@ -0,0 +1,76 @@
+import json
+import sys
+import pandas as pd
+from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score, confusion_matrix
+
+
+REQUIRED_PRED_COLUMNS = {"scenario_id", "prediction"}
+REQUIRED_TRUTH_COLUMNS = {"scenario_id", "label_future_collapse"}
+
+
+def validate_columns(df, required, name):
+    missing = required - set(df.columns)
+    if missing:
+        raise ValueError(f"{name} missing columns: {missing}")
+
+
+def validate_ids(df, name):
+    if df["scenario_id"].isna().any():
+        raise ValueError(f"{name} contains missing scenario_id")
+
+    if df["scenario_id"].duplicated().any():
+        raise ValueError(f"{name} contains duplicate scenario_id values")
+
+
+def validate_predictions(df):
+    if not df["prediction"].isin([0,1]).all():
+        raise ValueError("Predictions must be binary (0 or 1)")
+
+
+def main():
+
+    if len(sys.argv) != 3:
+        raise ValueError("Usage: python scorer.py predictions.csv truth.csv")
+
+    pred_path = sys.argv[1]
+    truth_path = sys.argv[2]
+
+    pred = pd.read_csv(pred_path)
+    truth = pd.read_csv(truth_path)
+
+    validate_columns(pred, REQUIRED_PRED_COLUMNS, "prediction file")
+    validate_columns(truth, REQUIRED_TRUTH_COLUMNS, "truth file")
+
+    validate_ids(pred, "prediction file")
+    validate_ids(truth, "truth file")
+
+    validate_predictions(pred)
+
+    merged = pred.merge(truth, on="scenario_id", how="inner")
+
+    y_true = merged["label_future_collapse"]
+    y_pred = merged["prediction"]
+
+    acc = accuracy_score(y_true, y_pred)
+    prec = precision_score(y_true, y_pred, zero_division=0)
+    rec = recall_score(y_true, y_pred, zero_division=0)
+    f1 = f1_score(y_true, y_pred, zero_division=0)
+
+    tn, fp, fn, tp = confusion_matrix(y_true, y_pred).ravel()
+
+    results = {
+        "accuracy": acc,
+        "precision": prec,
+        "recall_collapse": rec,
+        "f1": f1,
+        "true_negatives": int(tn),
+        "false_positives": int(fp),
+        "false_negatives": int(fn),
+        "true_positives": int(tp)
+    }
+
+    print(json.dumps(results, indent=2))
+
+
+if __name__ == "__main__":
+    main()