Streamlining Spare Part Management in Car Demolition Companies: A Web-Based Solution with Data-Driven Inventory and Sales Analysis

The automotive dismantling sector faces significant challenges in spare part inventory management, driven by irregular sales patterns, category variability, and multi-channel sales operations. This thesis presents an integrated software and data-driven solution to optimize spare part management for dismantling companies. The developed system combines a Flutter-based mobile and web applica- tion with FastAPI/Django middleware and an ODOO ERP backend, providing centralized control over inventory and seamless integration with external e- commerce platforms such as PrestaShop, eBay, and Subito.it. A dedicated data engineering pipeline merges, cleans, and aggregates sales, catalog, and stock data into consistent analytical datasets. On the analytical side, a probabilistic modeling framework was applied to characterize spare part sales behavior. Poisson, Negative Binomial, and Geo- metric distributions were fitted to daily, weekly, and monthly sales data, both globally and at the product category level. The models were validated through visual (histograms, CDF, survival functions) and quantitative (RMSE) evalua- tion. Empirical results revealed substantial overdispersion in sales data, rendering the Poisson model insufficient for accurate forecasting. The Negative Binomial model consistently provided superior fit, particularly for daily aggregated data, making it a robust baseline for demand forecasting. Geometric modeling further offered interpretable insights into short-term sales probabilities. The system was successfully deployed and tested in a real dismantling com- pany, demonstrating functional reliability and measurable improvements in in- ventory performance, including increased fill rate, reduced restocking delays, and minimized synchronization conflicts.