Integrating Machine Learning Based Sales Forecasting With Odoo Erp for Automated Inventory Management in a Retail Company

Efficient inventory management is crucial in the retail industry to ensure product availability, minimize operational costs, and enhance customer satisfaction. Conventional inventory systems often depend on manual processes and isolated forecasting techniques, which lead to issues such as inaccurate demand prediction, overstocking, stockouts, and delayed response to market fluctuations. These limitations underscore the necessity for an automated, predictive inventory management system seamlessly integrated within enterprise resource planning (ERP) workflows. This study presents an end-to-end solution integrating machine learning-based sales forecasting using the XGBoost algorithm with the Odoo ERP system. Historical sales data from January 2024 to January 2025 were utilized to train and validate the forecasting model. The model was deployed via FastAPI to deliver real-time predictions accessible through a custom Odoo module. Automated scheduled tasks enable weekly data exchange between the ERP and forecasting API, facilitating automatic inventory replenishment decisions. PostgreSQL serves as the backend database to maintain data integrity and performance. The research follows the Design Science Research Methodology (DSRM) to guide the iterative design, development, and evaluation process. Evaluation results indicate that the integrated system significantly improves inventory turnover by over 130 %, reduces total operational costs by 8.51 %, and maintains acceptable stockout levels, collectively enhancing product availability and customer satisfaction. This research constitutes the first known implementation in Indonesia of automated machine learning forecasting embedded directly within ERP workflows, providing a scalable framework for advancing retail inventory management.