Create preprocessing.py

utils/preprocessing.py

@@ -0,0 +1,23 @@
+import pandas as pd
+from sklearn.ensemble import IsolationForest
+
+def clean_data(file):
+    """
+    Bersihkan data UMKM dari anomaly (outlier).
+    Contoh input: File CSV dengan kolom: tanggal, demand, supply
+    """
+    # Baca data
+    df = pd.read_csv(file)
+    
+    # Konversi tanggal
+    df['tanggal'] = pd.to_datetime(df['tanggal'])
+    
+    # Deteksi anomaly
+    clf = IsolationForest(contamination=0.05, random_state=42)
+    df['anomaly'] = clf.fit_predict(df[['demand', 'supply']])
+    
+    # Filter data bersih
+    clean_df = df[df['anomaly'] == 1].copy()
+    clean_df.drop('anomaly', axis=1, inplace=True)
+    
+    return clean_df