import pandas as pd
from sklearn.ensemble import IsolationForest

def clean_data(file):
    # Baca dan format data
    df = pd.read_csv(file)
    df['tanggal'] = pd.to_datetime(df['tanggal'])
    df = df.sort_values('tanggal').reset_index(drop=True)
    
    # Deteksi anomali
    clf = IsolationForest(contamination=0.05, random_state=42)
    df['anomali'] = clf.fit_predict(df[['demand', 'supply']])
    
    return df[df['anomali'] == 1].drop('anomali', axis=1)