Development of model for forecasting SKU demand based on Winter s approach

Abstract

Forecasting is the art and science of estimating unknown situations. Managers use forecasting tools extensively in order to arrive at reasonable future scenarios during the decision making. Thus forecasting is inherently related with the decision making and hence use of proper forecasting technique is important for any organization. In this project we have studied and tried to develop a forecasting tool based on the Winter s method of forecasting an effective technique used for the data showing seasonal variations. A paints company has been using the same technique for the past few years to forecast the sales of its SKUs. It has been using a demand planner tool for the purpose. The tool and the method have given good results in the pasts, but these could be improved upon. Our objective of the study was to develop a model for the company such that the forecasting errors could be reduced. We studied the existing technique used by the demand planner tool and tried to find out the loopholes in the tool which were resulting in larger errors. For this we developed our own model on Excel which could be replicated on the demand planner. By introducing certain new methods in terms of parameter and model selection, we were able to achieve reduction in the errors. This report describes our methodology in achieving the desired results. During the process we also discovered that the results are not fully explicable mathematically but there are also some organizational and behavioral issues involved in using forecasting techniques in an organizational setting.