Rifai, Ahmad (2023) Analisis Atribut Profil Nasabah Sebagai Faktor Kelancaran Produk Ritel Kredit Perbankan Pada Bank XYZ. Masters thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 6032202022-Master_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
6032202022-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2025. Download (3MB) | Request a copy |
Abstract
Banyak lembaga yang menyediakan layanan kredit retail, mulai dari bank, financial technology (fintech), bahkan lembaga kredit yang diawasi OJK. Namun bank yang memberikan bunga cukup rendah dibandingkan dari lembaga pemberi kredit yang lain. Setiap bank memiliki naluri untuk meminimalisir risiko kemungkinan nasabah gagal bayar. Namun, dengan meningkatnya realisasi penyaluran kredit membuat pihak bank harus memperketat protokol dalam pengajuan kredit. Sehingga sangat penting untuk pihak bank mengklasifikasi nasabah yang mengajukan kredit agar tidak terjadi default dan sebagai protokol utama dalam memilah para peminjam. Sehingga peneliti ingin mengklasifikasi kolektabilitas nasabah berdasarkan profil dan informasi keuangan nasabah dengan menggunakan beberapa metode klasifikasi menggunakan machine learning, yaitu naïve bayes menggunakan SMOTE, Catboost, Light Gradient Boosting Machine, dan random Forest menggunakan SMOTE. Variabel yang digunakan pada analisis machine learning sebanyak 26 variabel. Berdasarkan hasil analisis, diketahui bahwa metode LGBM yang memiliki skor AUC dan skor feature importance tertinggi. Sehingga Metode LGBM digunakan pada penelitian ini. Berdasarkan variabel-variabel yang termasuk dalam fitur paling penting diketahui dapat menunjukkan beberapa perbedaan antar Legacy berdasarkan variabel yang termasuk feature importance untuk menentukan kebijakan yang tepat dan menindaklanjuti perbedaan tersebut.
================================================================================================================================
Many institutions provide retail credit services, including banks, financial technology (fintech), and even credit institutions supervised by the OJK. However, banks that provide interest are pretty low compared to other lending institutions. Every bank has the instinct to minimize the risk of possible default by customers. However, with the increase in the realization of credit distribution, the bank had to tighten the protocol for applying for credit. The bank needs to classify customers who apply for credit so that default does not occur and as the main protocol in sorting out borrowers. Researcher want to classify customer collectability based on customer profiles and financial information using several classification methods using machine learning, namely naïve Bayes with SMOTE, Catboost, Light Gradient Boosting Machine, and Random Forest with SMOTE. The variables used in machine learning analysis are 26 variables. Based on the results of the research, it is known that the LGBM method has the highest AUC score and feature importance score. Hence, the LGBM method was used in this study. Based on the variables included in the most important features it is known that it can show some differences between Legacy based on variables including feature importance to determine the right policy and follow up on these differences.
Actions (login required)
 |
View Item |