ANALISIS POTENSI KEPATUHAN TERHADAP PEMBAYARAN PAJAK BUMI DAN BANGUNAN PERDESAAN DAN PERKOTAAN (PBB-P2) MENGGUNAKAN ALGORITMA C4.5 DAN CLASSIFICATION AND REGRESSION TREES (CART) (STUDI KASUS: BADAN PENDAPATAN DAERAH KOTA TANGERANG)

Abstract

Land and Building Tax for Rural and Urban Areas (PBB-P2) is one of the main sources of local revenue in Tangerang City. Taxpayer compliance in paying PBB-P2 plays an important role in optimizing regional tax revenue. This study aims to analyze taxpayer compliance potential and compare the performance of the C4.5 and Classification and Regression Trees (CART) algorithms in predicting PBB-P2 payment status. The research data were obtained from the Regional Revenue Agency (Bapenda) of Tangerang City with a total of 10000 taxpayer records. The variables used include BUKU, Land NJOP Class, Building NJOP Class, Total NJOP, Land Area, Building Area, District, and Payment Status. The research method involved data preprocessing, encoding, splitting data into training and testing sets, and classification using Decision Tree algorithms implemented in Python. The results showed that the C4.5 algorithm achieved an accuracy of 85,70%, while the CART algorithm achieved an accuracy of 86,10%. In addition, both algorithms were able to identify the most influential attributes affecting taxpayer compliance, namely Building NJOP Class, BUKU, and Land NJOP Class. Based on the evaluation results, the CART algorithm performed better than the C4.5 algorithm.