10. BUSINESS ELEMENT FORECASTING › 10.4 Analytic Technique Tutorial › 10.4.1 Developing Business Element Models

10.4.1 Developing Business Element Models

You can develop forecasting models based on business elements
using a seven-step process.  Because of the significant
amount of time required to interview the users and developers
of each application that you wish to model, we recommend that
you reserve business element-based forecasting for
significant applications where Univariate Model Forecasting
has proved unsuccessful (that is, one or more forecasts
failed to track the actual values).  The seven development
steps are:

 1.  Interview the developers of the application to determine
     what they believe are the most significant influences on
     the execution characteristics of the application (that
     is, data volume associated with various input record
     types or input volume of various files).

     While this interview probably will not identify actual
     business elements, you must have a good understanding of
     the application's characteristics before you interview
     the application's users.

 2.  Interview the users of the application to determine the
     principal volume characteristics of their function.  For
     example, the principal volume characteristics of a
     department in an insurance company might be new
     policies, renewals, and claims.  You can use the
     knowledge of the application that you obtained from the
     developers to make initial guesses as to which of these
     business elements has the greatest correlation to the
     resource consumption of the application you are
     modeling.

 3.  Obtain historical observations for the business elements
     identified in the interview.  In the insurance example
     discussed above, you would want to obtain monthly or
     weekly historical values (depending on the timespan for
     which you wanted to model the application) for the
     number of new policies, renewals, and claims.  You would
     also obtain future estimates of these business volumes
     from the users.

 4.  Create a business element history file in the capacity
     planning database that contains the historical and
     future estimates of the business elements provided by
     the users.

 5.  Identify a resource element history file that tracks the
     computer resources used by the application.  Create such
     a file if it does not already exist.  Depending upon the
     way the installation defines performance groups,
     reporting performance groups, job names, and program
     names, you might choose to build this file from the
     CA MICS SCPPGA, SCP_RG, BATJOB, and/or BATPGM Files.

 6.  Use Business Element Forecasting to select business
     elements that have a high correlation to the resource
     consumption of the application.  Use the future
     estimates of the selected business elements to develop
     forecasts of the future resource consumption values.
     Note that the future estimates may be provided by you or
     generated by one of the forecasting routines of this
     component.

 7.  Continue to track the forecast of the application's
     resource consumption and the estimates of the business
     elements provided by the application's users.  Be aware
     that accuracy of this methodology is dependent not only
     on the characteristics of the statistical model
     proposed, but the accuracy of the business element
     estimates provided by the application's users.