PROCUREMENT FORECASTING IN ENTERPRISE RESOURCE AND PLANNING (ERP) SYSTEM USING HYBRID TIME SERIES MODEL

Abstract

Forecasting is one of the useful applications of machine learning in Enterprise Resource Planning (ERP). This is a vital component of a business's financial planning, which projects into the future using data from the past. This study offers a novel method for anticipating purchases using time series data. Based on prior order requests, the suggested approach is utilized to determine the buy price for specified items for possible purchase orders from suppliers. The dataset used during implementation is based on a real company purchasing process extracted from the SAP Business One ERP system. The dataset contains 12 years of historical monthly procurement data. It covers the period from 2010 to 2022, having a total number of 7600 observations. The evaluation of the proposed approach is measured using the Mean Absolute Error (MAE) and Root Mean Squared Error (RSME), respectively.