Web-Based Inventory Management System for Educational Training: Integrating EOQ and ARIMA for Data-Driven Learning

Manual inventory recording and heuristic ordering practices remain common among Micro, Small, and Medium Enterprises (MSMEs), often leading to inaccurate demand estimation, excessive holding costs, and stockouts. This study develops and evaluates a web-based inventory information system that integrates Autoregressive Integrated Moving Average (ARIMA) forecasting with the Economic Order Quantity (EOQ) model to improve decision accuracy and cost efficiency. The system uses CodeIgniter 3 and MySQL and incorporates a Python-based time-series forecasting engine. Historical sales data were modeled using ARIMA, and the optimal specification was selected based on Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC). The ARIMA(1,1,1) model achieved a Mean Absolute Percentage Error (MAPE) of 8.47%, indicating high forecasting accuracy for operational planning. The forecasted annual demand was integrated into the EOQ framework to determine the optimal order quantity, Reorder Point (ROP), and probabilistic Safety Stock. A one-year cost simulation demonstrated that the EOQ-based policy reduced total inventory costs by 22.73% compared with the existing approach. Functional validation through Black-Box testing confirmed full compliance with specified requirements. These findings demonstrate that integrating predictive analytics with classical inventory optimization enhances operational efficiency and reduces total inventory cost. The system provides a practical, data-driven inventory management framework for MSMEs undergoing digital transformation.