Supply Chain Demand Forecasting based on Odoo ERP System, Using Linear Regression and Decision Trees

Abstract

The transactional data coming from a software system’s transactional database provides more value than the day-to-day operations. Data mining techniques may be used to draw more value from these transactional data in order to enhance decision making process. Demand Forecasting is an undoubtedly essential strategic tool to any profit-seeking organization who are seeking to decrease their operational costs. This study conducted a conjunction between CRISP-DM process and Ralph Kimball’s data warehouse dimensional modelling methodology; in order to forecast sales demand in a given time frame. The application of CRSIP-DM went through two consecutive investigations. The first model used the Multiple Linear Regression which showed limitations in the model due to the mixed nature of ERP’s data. This has led to the second model where Decision Tree C4.5 was used. The forecasting model showed remarkable accuracy in the forecasting of the sales demand.