Identifikasi Potensi Penipuan pada Transaksi Bank Menggunakan K-Means Clustering

Abstract

Increasing cases of fraud in bank transactions are a serious concern for financial institutions, resulting in significant economic losses and undermining customer trust. This calls for identifying suspicious transaction patterns through machine-learning approaches to mitigate the risk of fraud. The methods used include problem identification, transaction data collection, and preprocessing to clean and prepare the data and after that, applying the K-Means Clustering algorithm to group transactions based on similar characteristics. The evaluation result obtained in this study using the Silhouette Score is 0.42, indicating a fairly good separation between normal and suspicious transactions. This research is expected to contribute to the development of a more accurate and efficient machine learning-based fraud detection system in banking institutions.