Analisis Kolektibilitas Nasabah Bank XYZ Menggunakan Random Forest Dan Xgboost

Putri, Natasya Liana (2026) Analisis Kolektibilitas Nasabah Bank XYZ Menggunakan Random Forest Dan Xgboost. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2043221010-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Dalam mendukung inklusi keuangan dan menjaga stabilitas ekonomi nasional, bank memiliki tanggung jawab penting dalam mengelola pembiayaan, khususnya terkait penyaluran kredit. Bank XYZ sebagai lembaga keuangan menghadapi permasalahan adanya peningkatan eksposur risiko kredit nasabah Malang sehingga munculnya potensi gagal bayar karena pembayaran kredit yang tidak lancar. Sehingga Bank XYZ perlu menerapkan strategi berbasis data untuk mengidentifikasi pola kolektibilitas nasabah dalam memprediksi pola pembayaran nasabah. Hal ini untuk mengantisipasi risiko kredit berupa kerugian finansial bagi pihak bank karena dana yang telah disalurkan tidak dapat kembali sepenuhnya yang mungkin terjadi di masa depan sehingga berdampak pada penurunan total pendapatan institusi keuangan tersebut. Penelitian ini bertujuan untuk menganalisis tingkat kolektibilitas nasabah menggunakan pendekatan machine learning dengan memberikan rekomendasi metode terbaik diantara XGBoost dan Random Forest dan informasi variabel yang paling berkontribusi besar dalam menentukan ketidaklancaran pembayaran nasabah. Dalam menangani overfitting dan ketidakseimbangan data dilakukan pengambilan sampel dengan Random Undersampling. Penelitian ini menunjukkan bahwa metode terbaik dalam analisis kolektibilitas nasabah Bank XYZ adalah XGBoost setelah dihipertuning menggunakan RandomSearchCV dengan parameter terbaik adalah learning rate=0,1, max_depth=7, n_estimators=1000, sub_sample 0,8 serta min_child_weight=1 yang mampu menghasilkan akurasi 86,27%, precision 83,50%, recall 90,37%, f1-score 86,80%, dan AUC 92,66%. Berdasarkan hasil SHAP diperoleh besar pinjaman menjadi faktor utama yang berkontribusi atas ketidaklancaran pembayaran nasabah. Dimana pinjaman besar, margin tinggi, tenor pendek, jenis pekerjaan wiraswasta, dan akad murabahah atau rahn cenderung memiliki kolektibilitas tidak lancar.
===============================================================================================================================
In supporting financial inclusion and maintaining national economic stability, banks have an important responsibility in managing financing, particularly with regard to credit distribution. Bank XYZ, as a financial institution, faces the issue of increased credit risk exposure among customers in Malang, which has led to potential default due to non-performing loan repayments. This condition indicates that Bank XYZ needs to implement data-driven strategies to identify customer collectability patterns in order to predict repayment behavior. Such efforts are intended to anticipate credit risk in the form of financial losses for the bank, as disbursed funds may not be fully recovered in the future, which could ultimately lead to a decline in the financial institution’s total revenue. This study aims to analyze customer collectability levels using a machine learning approach by providing recommendations on the best-performing method between XGBoost and Random Forest, as well as identifying the variables that contribute most significantly to customer payment delinquency. To address overfitting and data imbalance, a Random Undersampling technique is applied. The results show that the best method for analyzing customer collectability at Bank XYZ is the XGBoost model after hyperparameter tuning using RandomSearchCV with best parameters learning rate=0,1, max_depth=7, n_estimators=1000, sub_sample 0,8 serta min_child_weight=1. This model achieved an accuracy of 86,27%, precision of 83,50%, recall of 90,37%, F1-score of 86,80%, and a ROC curve value of 92,66%. Based on SHAP results, loans is identified as the primary factor contributing to customer payment delinquency, where large loans, high margins, short tenors, self-employment, and murabahah or rahn contracts tend to result in poor collectibility

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Bank, Kolektibilitas, Random Forest, XGBoost, SHAP, Bank, Collectibility, Random Forest, XGBoost, SHAP.
Subjects:	H Social Sciences > HG Finance > HG3751 Credit--Management. Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions:	Faculty of Vocational > 49501-Business Statistics
Depositing User:	Natasya Liana Putri
Date Deposited:	30 Jan 2026 06:40
Last Modified:	30 Jan 2026 06:40
URI:	http://repository.its.ac.id/id/eprint/131287

Actions (login required)

View Item