Enhancing Demand Forecasting Accuracy for Retail Equipment A Quantitative Assessment of Forecasting Methods

Abstract

In a globalized world with supply chains becoming more and more complex, the need of being more accurate in predicting future sales becomes essential for com- panies. In this thesis, the focal company was IKEA Components and the demand forecasting process of the retail equipment was analyzed. The thesis aimed to an- swer the question of what are the most suitable forecasting methods and parameters to accurately predict the demand for the retail equipment at IKEA Components. The current method used by IKEA Components was compared with 7 proposed methods and the results showed an accuracy improvement in 87% of the retail equipment stock keeping units (SKU). For the majority of SKUs with smooth de- mand, the best-performing forecasting method is Moving Average (MA) followed by Double Exponential Smoothing (DES) due to the low fluctuations in demand. For items with erratic demand, the Triple Exponential Smoothing (TES) Additive method stands out for its ability to closely follow the sharp fluctuations in demand. Each suggested method yielding the smallest error for at least one SKU, proved that an one size fits all approach is not suitable for the retail equipment at IKEA Components. As a result from the selection of the best-performing forecasting methods, some in- sights were found. First, demand patterns were identified and the one that involves the largest portion of SKUs are trend and seasonality occurring simultaneously. Second, correlation in demand behavior between subcategories of SKUs were iden- tified. Furthermore, suggestions on how to further enhance forecast accuracy by applying qualitative assessment were given. Finally, it was theoretically stated that there are positive economic, environmental and social implications of improving the forecasting process.