ANALISIS PERBANDINGAN ALGORITMA K-NEAREST NEIGHBOR (KNN) DAN SUPPORT VECTOR MACHINE (SVM) DALAM KLASIFIKASI DATA PERBANKAN

Abstract

Banking data is characterized by its diverse nature, encompassing financial transactions, credit histories, customer details, and more. This complexity poses challenges in effective data management and analysis. One such challenge lies in assessing the creditworthiness of home loan applicants. In the lending process, several factors are considered, including the applicant's character, repayment capacity, collateral, economic condition, age, and funding source. Credit risk analysis can be undertaken using various methods, including market analysis and big data machine learning techniques. In this study, we employed two popular algorithms: K-Nearest Neighbors (KNN) and Support Vector Machine (SVM). KNN, known for its simplicity and ability to work with diverse data, is frequently used for classification. However, it faces limitations in terms of computational speed on large datasets and susceptibility to noisy data. Conversely, SVM, a more sophisticated algorithm, is designed to maximize the separation between data classes and excels in handling high-dimensional data. Our findings reveal that SVM outperforms KNN in classifying credit risk in banking data, particularly for home loans. SVM with dot and ANOVA kernels achieved an accuracy of 80.56%, compared to 69.44% for KNN with k=50. These findings indicate that SVM is superior in credit risk classification in banking data, especially for home ownership loans.