MAHENDRA, RAFFI DZAKY (2026) OPTIMIZING RANDOM FOREST FOR MULTICLASS CREDIT SCORING CLASSIFICATION WITH SMOTE AND FEATURE SELECTION. S1 thesis, UNIVERSITAS KHATOLIK SOEGIJAPRANATA.

Preview	Text 22.K3.0003 - RAFFI DZAKY MAHENDRA-COVER_1.pdf Download (338kB) | Preview
	Text 22.K3.0003 - RAFFI DZAKY MAHENDRA-ISI_1.pdf Restricted to Registered users only Download (1MB)
Preview	Text 22.K3.0003 - RAFFI DZAKY MAHENDRA-DAPUS _1.pdf Download (271kB) | Preview
	Text 22.K3.0003 - RAFFI DZAKY MAHENDRA-LAMP _1.pdf Restricted to Registered users only Download (351kB)

Abstract

Credit scoring is the process of evaluation and interpretation stages carried out by financial services institutions to determine a person’s creditworthiness based on their financial history data. The main challenge in conducting credit scoring classification lies in the large number of interrelated features and the imbalance in class distribution. Most previous studies have only focused on improving algorithm performance, applying feature selection methods, and applying balancing data methods separately so that the effectiveness of the combination of these approaches, especially in the context of multiclass classification has not received adequate attention. This study attempts to fill this gap by optimizing the Random Forest algorithm through the application of the SMOTE method and two feature selection methods, namely Recursive Feature Elimination (RFE) and Mutual Information. A total of six modeling scenarios were used to evaluate the effect of using the feature selection and balancing data methods with the Random Forest algorithm to identify the best scenario in performing multiclass credit scoring classification. Model evaluation was carried out using various evaluation metrics such as accuracy, macro precision, macro recall, and macro F1-Score. The results of this study showed that the use of RFE provided the best performance with an accuracy of 83.21%, macro precision of 82.24%, macro recall of 82.86%, and macro F1-Score of 82.53%.

Actions (login required)

View Item