scorer.py

import sys
import json
import pandas as pd
from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score, confusion_matrix

def load_csv(path):
    try:
        return pd.read_csv(path)
    except Exception as e:
        print(f"Error loading {path}: {e}")
        sys.exit(1)

def main():
    if len(sys.argv) != 3:
        print("Usage: python scorer.py predictions.csv data/test.csv")
        sys.exit(1)

    pred_path = sys.argv[1]
    truth_path = sys.argv[2]

    pred = load_csv(pred_path)
    truth = load_csv(truth_path)

    required_pred = {"scenario_id", "prediction"}
    if not required_pred.issubset(pred.columns):
        print("Error: predictions.csv must contain scenario_id,prediction")
        sys.exit(1)

    if "scenario_id" not in truth.columns or "label" not in truth.columns:
        print("Error: test.csv must contain scenario_id,label")
        sys.exit(1)

    try:
        pred["prediction"] = pred["prediction"].astype(float).astype(int)
        truth["label"] = truth["label"].astype(float).astype(int)
    except (ValueError, TypeError):
        print("Error: prediction and label columns must contain 0 or 1.")
        sys.exit(1)

    merged = truth.merge(pred[["scenario_id", "prediction"]], on="scenario_id", how="inner")

    if len(merged) == 0:
        print("Error: no matching scenario_id values.")
        sys.exit(1)

    y_true = merged["label"]
    y_pred = merged["prediction"]

    cm = confusion_matrix(y_true, y_pred, labels=[0, 1])

    results = {
        "accuracy": accuracy_score(y_true, y_pred),
        "precision": precision_score(y_true, y_pred, zero_division=0),
        "recall": recall_score(y_true, y_pred, zero_division=0),
        "f1": f1_score(y_true, y_pred, zero_division=0),
        "confusion_matrix": {
            "tn": int(cm[0][0]),
            "fp": int(cm[0][1]),
            "fn": int(cm[1][0]),
            "tp": int(cm[1][1])
        }
    }

    print(json.dumps(results, indent=2))

if __name__ == "__main__":
    main()