Forecasts are vital to every business organization and for every significant management decision. While a forecast is never perfect due to the dynamic nature of the external business environment, it is beneficial for all levels of functional planning, strategic planning, and budgetary planning. Decision-makers use forecasts to make many important decisions regarding the future direction of the organization.
Forecasting techniques and models can be both qualitative and quantitative, and their level of sophistication depends on the type of information and the impact of the decision. The forecasting model a firm should adopt depends on several factors including: forecasting time horizon, data availability, accuracy required, size of the forecasting budget, and availability of qualified personnel.
Demand management exists to coordinate and control all sources of demand so the productive system can be used efficiently and the product delivered on time. Demand can be either dependent on the demand for other products or services or independent because it cannot be derived directly from that of other products.
Forecasting can be classified into four basic types: qualitative, time series analysis, causal relationships, and simulation. Qualitative techniques in forecasting can include grass roots forecasting, market research, panel consensus, historical analogy, and the Delphi method. Time series forecasting models try to predict the future based on past data. A simple moving average forecast is used when the demand for a product or service is constant without any seasonal variations. A weighted moving average forecast varies the weights, given a particular factor and is thus able to vary the effects between current and past data.
Exponential smoothing improves on the simple and the weighted moving average forecasts since exponential smoothing considers the more recent data points to be more important. To correct for any upward or downward trend in data collected over time periods to smoothing constants are used. Alpha is the smoothing constant, while delta reduces the impact of the error that occurs between the actual and the forecast.
Forecast errors are the difference between the forecast value and what actually occurred. All forecasts contain some degree of error; however, it is important to distinguish between sources of error and measurement of error. Sources of error are random errors and bias. Various measurements exist to describe the degree of error in a forecast. Bias errors occur when a mistake is made, i.e., not including the correct variable or shifting the seasonal demand. While random errors cannot be detected, they occur normally.
A tracking signal indicates whether the forecast average is keeping pace with any movement changes in demand. The MAD or the mean absolute deviation also is a simple and useful tool in obtaining tracking signals. A more sophisticated forecasting tool to define the functional relationship between two or more correlated variables is linear regression. This can be used to predict one variable given the value for another. It is useful for shorter time periods as it assumes a linear relationship between variables.
Causal relationship forecasting attempts to determine the occurrence of one event based on the occurrence of another event. Focus forecasting tries several rules that seem logical and easy to understand to project past data into the future. Today many computer forecasting programs are available to easily forecast variables. When making long-term decisions based on future forecasts, great care should be taken to develop the forecast. Likewise, multiple approaches to forecasting should be used.