Amiri, Yusril Izzi Arlisa (2020) Pemodelan Kemampuan Debitur Dalam Membayar Kredit di KUD X Menggunakan Bayesian Mixture Multinomial Regression. Masters thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
06211850010010-Master_Thesis.pdf Download (4MB) | Preview |
Abstract
Koperasi Unit Desa (KUD) X merupakan lembaga yang bergerak di bidang jasa keuangan. Salah satu kegiatan usaha dari KUD yaitu dapat menghimpun dana dan menyalurkannya melalui kegiatan usaha simpan pinjam. Dalam proses memberikan pinjaman, resiko kredit merupakan faktor penting yang perlu dikelola. Untuk memperkecil resiko gagal bayar, maka pihak koperasi dalam memberikan pinjaman perlu menerapkan prinsip kehati-hatian. Oleh karena itu, perlu dilakukan pemodelan sebagai upaya untuk memprediksi calon debitur yang tidak beresiko menyebabkan gagal bayar. Kemampuan membayar debitur dikategorikan menjadi 3, sehingga mengikuti distribusi multinomial. Debitur di KUD X terdiri dari 2
golongan debitur berdasarkan sistem pembayaran. Oleh karena itu, terbentuk struktur mixture dari distribusi multinomial dengan dua komponen. Struktur tersebut digunakan untuk menentukan kemampuan pemohon dalam membayar kredit dengan menggunakan Bayesian mixture multinomial regression model. Estimasi parameter menggunakan Bayesian Markov Chain Monte Carlo (MCMC) dengan algoritma Gibbs Sampling. Berdasarkan hasil analisis, diperoleh bahwa variabel jumlah pinjaman, tanggungan keluarga dan pendapatan merupakan variabel yang berpengaruh signifikan terhadap gagal bayar dibandingkan lancar untuk model non-mixture dan model mixture. Performa dari Bayesian mixture multinomial regression model diukur dengan menggunakan persentase klasifikasi dari kemampuan debitur dibandingkan dengan multinomial logistic regression
untuk model non-mixture menggunakan pendekatan Maximum Likelihood dan Bayesian. Hasil perbandingan menunjukkan bahwa Bayesian mixture multinomial regression memberikan performa dari nilai accuracy, precision, recall dan f-measure yang lebih tinggi daripada model non-mixture.
================================================================================================
Village Unit Cooperative (VUC) is an institution which provides financial services consisting of several sub-businesses, one of which is providing loans. To minimize the risk of non-performing loans, the cooperative must be able to assess prospective borrowers before a loan decision is taken. In this case, loan repayment with its three categories coupled with two type of installement and interest payment systems is constructed to produce a proposed structure of mixture multinomial distribution with two components. This structure is employed to determine loan repayment of the applicant by using the Bayesian mixture multinomial regression model. The estimating parameter is done by building an algorithm based on Bayesian Markov Chain Monte Carlo (MCMC) couple with the Gibbs Sampling algorithm. Based on the result of analysis, the loan amount, number of families and income are the significant variables for default both non-mixture model and mixture model. The classification performance of loan repayment through the Bayesian mixture multinomial regression model is measured by determining loan repayment classification percentage of the model which is compared with loan repayment classification percentage of the multinomial logistic regression for non-mixture model by using Maximum Likelihood and Bayesian. The comparative results show that Bayesian mixture multinomial regression model approach gives a higher percentage of accuracy, precision, recall, and f-measure of loan repayment
classification performance than non-mixture model.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | Bayesian Mixture Multinomial Regression, Bayesian Multinomial Logistic Regression, Bayesian MCMC, Gibbs Sampling, Multinomial Logistic Regression ===================================================================================== Bayesian Mixture Multinomial Regression, Bayesian Multinomial Logistic Regression, Bayesian MCMC, Gibbs Sampling, Multinomial Logistic Regression |
| Subjects: | H Social Sciences > HA Statistics > HA31.3 Regression. Correlation H Social Sciences > HA Statistics > HA31.7 Estimation |
| Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis |
| Depositing User: | Yusril Izzi Arlisa Amiri |
| Date Deposited: | 28 Aug 2020 01:37 |
| Last Modified: | 25 Dec 2023 12:58 |
| URI: | http://repository.its.ac.id/id/eprint/81468 |
Actions (login required)
 |
View Item |