Kasidi, Josua Christanto (2022) Perancangan Model untuk Prediksi Potensi Churn pada Debitur KPR dengan Regresi Logistik. Masters thesis, Insititut Teknologi Sepuluh Nopember.
Preview |
Text
6032201197-Master_Thesis.pdf - Accepted Version Download (1MB) | Preview |
Abstract
Kredit Kepemilikan Rumah (KPR) merupakan sistem pembiayaan dimana perbankan memberikan pinjaman kepada nasabah untuk mendapatkan rumah, dan melakukan pelunasan dalam waktu yang ditentukan. Oleh sebab itu, dalam persaingan bank di Indonesia, baik bank pemerintah maupun swasta kini berusaha untuk menawarkan program KPR kepada nasabah yang bertujuan agar nasabah merasa puas dengan pelayanan perbankan tersebut. Seiring dengan perjalanan umur KPR debitur, peristiwa churn customers (pelanggan yang meninggalkan perusahaan) dapat menjadi hal yang berdampak bagi pertumbuhan portfolio KPR serta membuat turunnya hubungan antara bank dengan nasabah.
Hingga kini, salah satu bank yang terus berusaha untuk mengatasi permasalahan churn customers adalah Bank Mandiri. Namun dalam menjawab tantangan tersebut terdapat beberapa permasalahan yang menjadi halangan untuk dapat menyelesaikannya. Salah satu permasalahan tersebut adalah kurangnya informasi mengenai faktor yang mempengaruhi churn customers akibat analisis yang terbatas karena jumlah perbandingan populasi yang tidak seimbang.. Oleh sebab itu, tujuan dari penelitian ini adalah untuk membuat model prediktif yang mampu melakukan klasifikasi terhadap nasabah churn pada produk KPR agar Bank Mandiri dapat melakukan retensi sedini mungkin.
Proses yang dilakukan untuk dapat mencapai tujuan tersebut dapat dimulai dengan mengelompokan kelompok nasabah menjadi data biner, dimana akan terdapat nasabah churn dan nasabah loyal. Kemudian penelitian dilanjutkan dengan mengumpulkan faktor-faktor yang terduga mempengaruhi nasabah churn, antara lain Recency, Frequency, dan Monetary (RFM), saldo tabungan, penghasilan, tagihan, Debt Burden Ratio (DBR), usia, rate, dan Debt Equity Ratio (DER), dan sisa plafon. Data yang digunakan pada penelitian ini adalah data sejarah transaksi, dana, dan kredit nasabah yang dimulai pada tahun 2019 – Mei 2021 yang terdiri dari 50,000 nasabah KPR. Data yang diperoleh akan dilanjutkan dengan melakukan pemodelan menggunakan regresi logistik biner. Untuk hasil pengujian, model yang digunakan akan diuji menggunakan nilai Receiver Operating Characteristic (ROC) dan Area Under Curve (AUC) sebagai nilai kelayakan dalam pengujian model data. Hasil akhir dari model ini diperoleh 7 variabel yang berpengaruh terhadap model disertai dengan nlai AUC dari logistik biner original sebesar 0.68 dan logistik biner SMOTE sebesar 0.72.
================================================================================================
A mortgage is a loan or financing system whereas banks provide loans to customers to get houses and make repayments within a specified time. Therefore, in banking competition in Indonesia, both state and private banks are now trying to offer mortgage to customers so that customers feel satisfied with banking services. The more customers who take mortgages will certainly have an impact on the quality of a bank's mortgage portfolio. Along with the age of mortgage loan, churn customers can have an impact on the growth of the mortgage portfolio and reduce the relationship between the bank and the customer.
Until now, one of the banks that continues to try to overcome the problem of churn customers is Bank Mandiri. However, in responding to these challenges, there are several problems that become obstacles to solve them. The lack of information about factor that affect churn customers still limited due to lack of information in unbalanced population. As a result, it will have an impact on mitigation actions that can be taken by banks before customer’s churn. Therefore, the purpose of this study is to create a predictive model that is able to classify churn customers on mortgage products so that Bank Mandiri can do retention just in time.
The process carried out to achieve this goal can be started by grouping customer groups into binary data, where there will be churn customers and loyal customers. Then the research continued by collecting factors that were suspected of influencing churn customers, including Recency, Frequency, and Monetary (RFM), savings balances, income, bills, Debt Burden Ratio (DBR), age, rate, Debt Equity Ratio (DER) and outstanding amount. The data used in this study is transaction history data, funds, and customer credit starting in 2019 – May 2021, consisting of 50,000 mortgage customers. The data obtained will be continued by modeling using binary logistic regression algorithm. For model validation, the model used will be tested using the Receiver Operating Characteristic (ROC) and Area Under Curve (AUC) values as the value of the feasibility of testing the data model. Then from the test results will be obtained a model to look for opportunities or predictive value and identification of the significant parameters that influence it based on the value of the important features. The final result of this model obtained 7 variabels that affect the model accompanied by the AUC value of the original binary logistic of 0.68 and the SMOTE binary logistic of 0.72.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | Kredit Kepemilikan Rumah, Churn customers, Predictive Analytics, Receiver Operating Characteristic, Area Under Curve, Sistem Peringatan Dini |
| Subjects: | H Social Sciences > HG Finance H Social Sciences > HG Finance > HG4012 Mathematical models T Technology > T Technology (General) > T11 Technical writing. Scientific Writing T Technology > T Technology (General) > T174 Technological forecasting T Technology > T Technology (General) > T57.8 Nonlinear programming. Support vector machine. Wavelets. Hidden Markov models. |
| Divisions: | Faculty of Creative Design and Digital Business (CREABIZ) > Technology Management > 61101-(S2) Master Thesis |
| Depositing User: | Josua Christanto |
| Date Deposited: | 24 Jan 2022 16:37 |
| Last Modified: | 29 Aug 2024 07:42 |
| URI: | http://repository.its.ac.id/id/eprint/92487 |
Actions (login required)
 |
View Item |