Optimizing Filling Shed Reconfiguration through K-Means Clustering with Silhouette, Association Rule Mining, and Artificial Neural Network

Abstract

Fuel oil is an essential resource in Indonesia, widely used in various sectors, including industry, transportation, and personal consumption. The company operates numerous branches, including integrated terminals throughout Indonesia, and is responsible for receiving products from refineries, storing them, and delivering them to consumers. The current system's inefficiencies obstruct the company's capacity to meet increasing demands, which could result in customer dissatisfaction and revenue loss. This study aims to propose a reconfiguration of the filling shed by segmenting, finding patterns and relationships between each transaction, and forecasting. The clustering result, by using 19 days of historical data, formed ten clusters, the association rules formed 248 rules, and the ANN showthat the model has an effective predicting ability. These results will be used to propose a reconfiguration that helps to overcome the bottleneck dealt with by the company.