Ökad tillförlitlighet i prognostisering av efterfrågan : En empirisk studie på fallföretaget Mycronic

Sammanfattning

A significant challenge within inventory management is balancing costs against maintaining a high service level for customers. Today, companies use forecasts to predict demand and thereby maintain a high service level without risking excessively high costs. However, if the forecast is not reliable, it can lead to poorly planned inventories. Major overforecasting can result in excess inventory, while major underforecasting can lead to stock shortages, which negatively affect the customer experience. Therefore, companies need to use reliable forecasts. Mycronic is a company that uses forecasts to plan its inventory management. However, the forecasts currently lack trust within the company because they often differ greatly from actual demand. As a consequence, inventory management becomes more difficult, leading to both stock shortages and overstock. According to Mycronic, the cause of this is seasonal variation in demand that the forecast does not take into account. To improve stock availability, it is therefore important to enhance the reliability of the forecast. Accordingly, the purpose of this study is formulated as follows: Identify improvement measures for Mycronic’s forecasting of demand to improve the realibility of the forecast. In the study, 29 articles, including both spare parts and accessories, were analysed. Deficiencies in forecasting and their consequences were identified through interviews with employees. Moreover, the current forecasting model within Mycronic’s ERP system was analysed, along with historical demand and forecasts for each article. By calculating the forecasting errors ME, MAD, and MAPE, it became evident that the current forecasting model was not optimal for all articles. Furthermore, deficiencies were identified beyond the choice of an appropriate forecasting model. For instance, there is no continuous monitoring, the forecast is sometimes based on inaccurate data, and knowledge about the forecasting process is lacking. When analysing the relationship between forecasting error and stock availability, it was shown in several cases that underforecasting correlates with stock shortages. This relationship demonstrates that stock availability can be improved by increasing the reliability of the forecast, in other words, by reducing the forecasting error. Based on the identified deficiencies, a number of forecasting models were selected to be tested on the articles. For each article, the forecasting model that resulted in the lowest forecasting error was chosen. The results showed that the forecasting model should preferably be adjusted according to the demand pattern of each article, for instance by accounting for trends or seasonal variations in order to generate more reliable forecasts. Furthermore, it became evident that the current forecasting model is too reactive, and that a more stable model would be better suited to the volatile demand. To address the identified deficiencies, a number of improvement measures were compiled into a list as follows: Implement processes for monitoring forecasts Adjust the forecasting model Decrease the reactivity of the forecast model Educate employees about forecasting Utilize AI/ML for forecasting Implement routine for to ensure correct input data Change method for calculating safety stock levels Update how the forecasts are shared with suppliers By implementing these improvement measures at Mycronic, their forecasting can improve, thereby reducing the risk of stock shortages and overstock. Additionally, the workload for employees will decrease as they become more confident that the forecast is reliable. Suppliers will receive a more reliable and stable forecast, enabling them to plan their organization better. It is thus relevant for companies other than Mycronic to improve their forecasts to enhance their operations. This study describes how companies with similar demand patterns for the articles as Mycronic can proceed to improve their forecasting and, consequently, the organization as a whole.