preprocessing/step3_class_balance_check.py

import pandas as pd

DATA_PATH = "dataset/processed/dataset_step2_cleaned.csv"

def main():
    df = pd.read_csv(DATA_PATH)

    # -------------------------------
    # Handle label column safely
    # -------------------------------
    if "Label" not in df.columns:
        if "Label (0- HUMAN, 1-AI)" in df.columns:
            df = df.rename(columns={"Label (0- HUMAN, 1-AI)": "Label"})
        else:
            raise ValueError("Label column not found in dataset.")

    # -------------------------------
    # Column validation
    # -------------------------------
    required_cols = ["Label", "Language"]
    missing = [c for c in required_cols if c not in df.columns]
    if missing:
        raise ValueError(f"Missing required columns: {missing}")

    total = len(df)

    print("\n📊 OVERALL CLASS BALANCE\n")
    class_counts = df["Label"].value_counts().sort_index()

    for label, count in class_counts.items():
        percent = (count / total) * 100
        class_name = "Human" if label == 0 else "AI"
        print(f"{class_name} ({label}): {count} samples ({percent:.2f}%)")

    print("\nTotal samples:", total)

    # -------------------------------
    # Language-wise breakdown
    # -------------------------------
    print("\n🌐 LANGUAGE-WISE CLASS DISTRIBUTION\n")

    for lang in df["Language"].unique():
        print(f"Language: {lang}")
        lang_df = df[df["Language"] == lang]
        lang_total = len(lang_df)

        lang_counts = lang_df["Label"].value_counts().sort_index()
        for label, count in lang_counts.items():
            percent = (count / lang_total) * 100
            class_name = "Human" if label == 0 else "AI"
            print(f"  {class_name} ({label}): {count} samples ({percent:.2f}%)")

        print(f"  Total {lang} samples: {lang_total}\n")


if __name__ == "__main__":
    main()