Scaling for E-Business

12.6. Forecasting Techniques

The literature describes several forecasting techniques [9] [10]. In selecting one of them, a few factors need to be considered. The first one is the availability and reliability of historical data. The accuracy and the planning horizon are also factors that determine the forecasting technique. The pattern found in historical data has a strong influence on the choice of the technique. The nature of historical data may be determined through visual inspection of a plot of the data as a function of time. Data collected in chronological order are called a time series. As displayed in Fig. 12.5, four patterns of historical data can be clearly identified: trend, cyclical, seasonal, and stationary. While the trend pattern ...