Sales Forecasting Using Exponential Smoothing

Sales forecasting affects almost every area of activity in industry. The importance of a sales forecast can never be underestimated. The choice of the right forecasting technique is essential for a company to operate efficiently. The purpose of this thesis is to show that exponential smoothing as a sales forecasting device and as a device to predict demand for production and inventory control purposes is more accurate, more efficient and less time consuming in its application than other conventional forecasting techniques. Exponential smoothing is a simple procedure for calculating a weighted moving average; the greatest weight is assigned to the most recent data of actual or predicted sales. This paper discusses the effectiveness of simple methods of exponential smoothing with regard to accuracy, computational simplicity, and flexibility in order to adjust the prediction to the rate of response of the forecasting system. It is not necessary to select and work with complicated economic indicators, etc. A number of commonly used forecasting devices are presented and an analysis of their strengths and weaknesses are discussed. Among the statistical forecasting tools constructively criticized are: correlation analysis (simple, multiple, linear and non linear), time series analysis, and moving averages (simple and weighted). Actual problem solutions and valid arguments are presented to prove that exponential smoothing is highly advantageous. Practical applications of exponential smoothing show real life cases where the experiences of a number of companies indicate exponential smoothing to be extremely beneficial. The academic discussions as well as practical applications of the technique in operation indicate exponential smoothing to be a most successful method. It requires less data than any type of forecasting method while remaining highly flexible because a modified forecast can be made by simply changing the smoothing constant. When used in conjunction with data processing equipment, exponential smoothing makes it possible to forecast demand accurately on a weekly basis. It is easily adapted to high speed electronic computers so that expected demand as well as detection of and correction for trends can be measured as a routine matter. It makes it possible to measure current distribution of forecast errors item by item. Therefore, exponential smoothing is particularly well suited for item forecasts which may be needed for determining re-order points, materials planning, economic order quantities in materials management and scheduling in production control