| Abstract [eng] |
Today when various social and economic factors change established consumer habits, demand forecasting is becoming one of the key tools that can help maintain or even lead in the retail market. A properly designed demand forecasting model provides an opportunity to improve the inventory management and marketing processes those are important to the company and those have a direct impact on the company’s financial results. There are many different ways and opinions on how to create the right model. The aim of this study is to analyze the different techniques using to create demand prediction models and based on the insights gained to develop a demand forecasting and inventory optimization model. The theoretical part of the work deals with research on inventory management systems, analytical programs and possible mathematical methods applied in different studies. The formed insights allow choosing fixed time ordering system, KNIME analytical program and Bayesian Additive Regression Trees (BART) mathematical method those will be used to develop a practical model for determining demand. In the research part, detailed sales data for January- June 2019 is selected, processed and adjusted at different stages so that the model developed using the Bayesian Additive Regression Trees mathematical method would be as accurate as possible. Based on available information and developed demand forecasting model, the need for replenishment is assessed on 11.07.2019. The results reveal that 399 products in ten different stores need replenishment because the available stock level cannot meet the projected demand. According to the author, replenishment model created could be improved by applying the economic order quantity model that with the optimal number of inventories defined may even better assist in company's inventory management processes. Thesis structure: introduction, three parts, conclusions, references and appendices. Thesis consists of 79 pages text without appendices, 10 tables, 15 pictures and 87 bibliographical entries. |
