KLASIFIKASI STATUS PENGAJUAN KPR RUMAH SEDERHANA MENGGUNAKAN ALGORITMA RANDOM FOREST

Abstract

The property industry has a strategic role in supporting economic growth, especially through Home Ownership Credit (KPR) financing which enables people to own housing. In big cities in Indonesia, various property developers are innovating to meet the demand for modern, affordable housing, but some of them are facing obstacles in the KPR feasibility evaluation process. Developers experience difficulties in managing the eligibility selection process, because inaccurate analysis can make consumers disappointed after paying the down payment (DP) if their KPR application is ultimately rejected. This research aims to classify the status of consumer mortgage applications, with historical data of 969 samples covering previous projects, which contains features such as consumer income, number of dependents, employment, credit history, and type of property being applied for, as well as 1 status feature which states the results of the application. The Random Forest algorithm is used to produce a KPR status classification model into three main categories: Contract, Reject Bank, and Reject Customer. Based on model evaluation using the average multiclass matrix, this model achieved 99% accuracy, with an average precision, recall and f1-score of 98%, which shows this model is very effective in identifying and classifying the three categories of KPR status.