14. WORKLOAD FORECASTING › 14.2 Usage Guidelines › 14.2.1 Neugents Technology Models

14.2.1 Neugents Technology Models

Neugents technology models utilize a nonlinear methodology to
first train using a set of client specified data elements and
then predict future values of the specified Forecast element.
The techniques employed are internally complex, but presented
in a simple, straight-forward manner, eliminating any
knowledge requirement of those processes.  This is in
contrast to normal statistical methods, which often deliver
satisfactory results, but generally require a considerable
amount of knowledge on the part of the client regarding
specific methods, measurements and output information.
Neugents technology provides accurate forecasting while
permitting the analyst to focus on the problem at hand, not
the underlying methodologies.

Aside from simplified use, Neugents technology are nonlinear
in nature, are generally less sensitive to "outliers" and
other data variability issues, and will normally produce more
accurate predictions using the types of data commonly found
in capacity planning applications.

The Resource Component Analysis application can be used to
evaluate the data elements being considered for training the
model. Training element values that show a Relative
Importance ranking (weight) value of less then 0.05 should
probably be dropped from training usage. Of course, the
analyst should always make such decisions based on his
personal knowledge of the resources and workloads, as well as
the business objectives of the data center.