Application of the K-Means Algorithm for Sales Data Clustering in the KartsPOS Application at PT Dizytech Bangun Asia

Abstract

The increasing availability of transaction data in point-of-sale applications creates opportunities to identify business patterns and support data-driven decision-making. This study aims to apply the K-Means clustering algorithm to group sales-related transaction data in the KartsPOS application at PT Dizytech Bangun Asia. The dataset consists of 925 transaction records collected over the observation period from October 2025 to October 2026. The research methodology includes data collection, data preprocessing, feature selection, determination of the number of clusters, K-Means implementation, and evaluation and interpretation of the clustering results. Two numerical attributes, namely POS/Webstore prices and invoice values, were selected and standardized before clustering. A total of 907 records with complete numerical attributes were processed using K-Means with three clusters. The monthly analysis revealed that September 2026 recorded the highest transaction volume, with 104 transactions, while August 2026 generated the highest invoice value of IDR 85,640,000. The clustering results produced three groups consisting of 807 low-value transactions, 88 medium-value transactions, and 12 high-value transactions. These findings demonstrate the potential of K-Means to organize transaction records according to their numerical characteristics and provide a structured overview of invoice-related activity. The resulting clusters can support transaction evaluation and business analysis at PT Dizytech Bangun Asia. However, the findings represent transaction segmentation rather than direct sales forecasting or profitability measurement.