Implementasi Algoritma Machine Learning untuk Pengembangan Model Prediktif atas Tingkat Non-Performing Loan dalam rangka Penjaminan Kredit UMKM Pemulihan Ekonomi Nasional

Reza Darmawan; Dwi Purnomo; Suryani  Nilawati; Fery Perdiansyah

Penulis

Reza Darmawan Kementerian keuangan
Dwi Purnomo Kementerian keuangan
Suryani Nilawati Kementerian keuangan
Fery Perdiansyah Kementerian keuangan

Kata Kunci:

Non-Performing Loan, SMEs Credit Guarantee for Economic Recovery Program, Machine Learning, Random Forest Regression

Abstrak

This analysis aims to develop a predictive model for SMEs’ credit NPL in order to support the decision making of Government SMEs Credit Guarantee program. The results of this analysis can be used to determine which SMEs business sector that is most affected by the COVID-19 pandemic, provide guarantee fees tariff suggestions that will be given to Jamkrindo and Askrindo as the guarantor institution for the National Economic Recovery program, as well as a tool for calculating the guarantee fees and loss limit budget. The predictive model was developed using a random forest regression algorithm using 2052 panel data for credit channeling to SMEs from the Indonesian Banking Statistics documents provided by, the Indonesian Financial Services Authority (OJK). The predictors used consist of the economic sector as well as macroeconomic conditions including inflation, unemployment, economic growth, and pandemic conditions.