13. RESOURCE COMPONENT ANALYSIS › 13.2 Analysis using CAP Database files › 13.2.1 Functional Description

13.2.1 Functional Description

Resource Component Analysis utilizes CA's U.S. patented
Neugents technology facility known as Relative Importance.
The Relative Importance facility utilizes proprietary
analytical technology to understand and quantify the effect
of a list of data values against the client specified
analysis data element.  While a similar analysis can be
performed using standard statistical means such as
correspondence matrix, the Relative Importance facility can
not only determine linear relationships among data values,
but non-linear relationships as well.  This is important when
exploring data that may not be well understood and can result
in a more accurate understanding of the relationships among
the data values.

As with other CA MICS Capacity Planner facilities, any files
to be analyzed using the Resource Component Analyzer must
reside in the capacity planning database.  The analysis can
be performed using either one or two database files, but both
files must be summarized in the same timespan and maintain
the same key structure.

Determine the data element to use for analysis and the
CA MICS Capacity Planner will produce a default list of all
quantitative data elements contained in the file defined
under "Select Elements from."  Edit this list as appropriate
by either selecting or unselecting data elements before
executing the application.

The application will validate all chosen independent elements
for 'missing' values. Such values are not valid for
performing Relative Importance and Neugents will ignore such
values during processing. Input observations containing
'missing' values will be bypassed and listed on an error
report.

Additionally, all training values will be analyzed to
determine if they are totally co-related with the forecast
value.  For example, in a database built from HARCPU data,
CPUTOBTM and CPUPCBSY would be totally co-related and as
such, CPUPCBSY values would bias the Relative Importance
process and cause the Importance Index (Weight) values to be
distorted. The application will compute co-relation weights
for all training elements and will cause any totally
co-related values to be dropped from analysis and flagged on
the Descriptive Statistics report.

During execution, the Relative Importance facility will
analyze the values of each selected element from the "Select
Elements from" list and develop a weight that quantifies the
effect of the data value against the Analysis element.  The
sum of the weights for all selected elements will be 100 and
will quantify the impact of each selected data element
against the analysis variable.  Those data values with a
greater weight have a more significant impact against the
analysis element than those with a lesser weight, and will
likely be of more interest in any future forecasting
operations.