Regression and Correlation Models:

Product demand can generally be linked to one or more causes (independent variables) in the form of an equation in which demand is the dependent variable. This kind of forecasting model may be developed using regression analysis. The usefulness of the regression equation is evaluated by the standard error of the estimate and the coefficient of determination r2. The first measures the expected uncertainty, or range of variation in a future forecast, while the second indicates the proportion of variation in demand explained by the independent variable(s) included in the model.

It is frequently advisable to start with a simple model that makes common sense and enrich it, if needed, for increased accuracy. Such an approach facilitates acceptance and implementation by management, whereas keeping the data collection and processing costs low.