10. BUSINESS ELEMENT FORECASTING › 10.4 Analytic Technique Tutorial

10.4 Analytic Technique Tutorial

Historically, one of the most difficult problems encountered
by the analyst was the forecasting of the resource
requirements of specific applications (ART79).  Perhaps the
most common approach to this problem is the use of periodic
user surveys to elicit estimates of future resource
requirements like EXCPs or CPU seconds.  Unfortunately, the
results of such surveys have been subject to significant
error since application users typically have little feel for
the relationship of their business activities to resource
consumption.  Therefore, investigators have proposed the use
of multiple regression techniques to build statistical models
that relate business activities to resource consumption
(ART79, SAR79, GOL79, BOW80).

For example, rather than asking the users of the payroll
application to estimate CPU consumption or other resource
requirements, you could attempt to build statistical models
based on factors like total employees, pay checks produced,
and/or other breakdowns of total employees such as
salary/hourly or weekly/bi-weekly/monthly pay periods.  These
factors have been given names such as natural forecasting
units, key volume indicators, and natural business units
(KOL76, SAR79, BRO78).  The CA MICS Capacity Planning
Component calls these factors business elements.

This section is an overview of the steps required to
establish a Business Element Forecasting model using CA MICS.
The primary analytical technique used in this process is
stepwise multiple regression.  The overview includes the
basic statistical concepts and terminology used in the
construction of the forecast.

The following topics are presented:

 1 - Developing Business Element Models
 2 - Stepwise Multiple Regression Concepts