9. MULTIVARIATE REGRESSION FORECASTING › 9.2 Usage Guidelines

9.2 Usage Guidelines

Multivariate Regression Forecasting is a technique for
investigating the mathematical relationships between one or
more resource variables contained in capacity planning data
base files.  The mathematical relationships which may exist
between variables with a high degree of correlation can be
expressed as models.  You can use these models to forecast
future values of the resource variables.

There are three broad steps in the development of a
multivariate regression model:

    o  Choosing model parameters
    o  Evaluating the model
    o  Forecasting future values from the model

These topics are discussed after a brief review of stepwise
multiple regression models.  This introduction presents some
of the basic terminology used in the evaluation of regression
models.  A more detailed development of the multiple
regression technique is presented in Section 9.4.

The following topics are presented by section:

    1 - Stepwise Multiple Regression Models
    2 - Choosing Model Parameters
    3 - Evaluating the Model
    4 - Forecasting Future Values from the Model