Optimizing Grocery Sales Data Grouping Using the Fuzzy C-Means Algorithm: Case Study of Nafhan Mart Store

Abstract

The sale of staple food products at Nafhanmart Store, Cirebon Regency, includes essential household items such as rice, cooking oil, sugar, and flour, which maintain stable demand as basic necessities. This study focuses on improving sales clustering models at Nafhanmart using the Fuzzy C-Means (FCM) algorithm, a prominent method in data mining. Key factors influencing sales include price, sales volume, demand, and remaining stock. Accurate clustering analysis is vital for strategic inventory management and profit maximization. The research applies the Knowledge Discovery in Database (KDD) methodology, encompassing data selection, preprocessing, transformation, FCM implementation, and evaluation using the Davies-Bouldin Index (DBI). Attributes analyzed include price, sales volume, demand, and remaining stock. The FCM algorithm clusters data based on patterns, with DBI evaluating clustering quality and determining optimal clusters. Data analysis and visualization were conducted using RapidMiner. Results show that the FCM algorithm achieves optimal clustering quality with a DBI score of 0.452 for two clusters, outperforming three clusters (DBI 0.474) and four clusters (DBI 0.536). Price and demand are identified as critical factors influencing clustering outcomes. These findings enhance the clustering model, offering actionable insights for inventory management and sales strategy, while showcasing the FCM algorithm's adaptability for other SMEs to support data-driven decision-making.