| ICEWARE ® ANALYTICS
ENGINE |
 |
In addition to searching data companies
are increasingly interested in finding trends and ‘rules’ within
their data. A common use of the resulting rules is in
one-to-one marketing campaigns. Many companies are restricted
in what they can achieve by the performance of the analysis
tools available. They tend to be relatively slow and
therefore will not work effectively over very large volumes
of data. This necessitates compromising on the size of
datasets analysed or one-pass rule analysis not allowing
any real validation of the resulting rules prior to using
them! The two issues mentioned above result in statistically
inaccurate and unproven rules. By using the Iceware ® Analytics
Engine within your application or by integrating it into
your current analytics toolset you can completely remove
those restrictions. The interface to the analytics engine
can be either a custom GUI or via software calls from
the application utilising it.
Many sectors can benefit from the Iceware ® Analytics
Engine, for example:
|
| Financial Services |
 |
Risk Management, Fraud, Marketing |
| Oil & Gas |
|
Analytics, Research |
| Retail |
|
Marketing |
| BioInformatics |
|
Fraud, Analytics |
| Biosciences |
|
Gene Analysis, Laboratory Analytics |
| Government |
|
Fraud, Marketing, Risk Management, eGovernment, Research
Establishments e.g Universities |
| Utilities |
|
Fraud, Analytics, Research Analysis |
| Telecommunications |
|
Location Based Services, Fraud, Marketing, Billing Analysis |
| Our current version of
the Analytics engine incorporates TDIDT. TDIDT (Top Down
Induction of Decision Trees) is a fundamental process used
in knowledge discovery and data mining. Using training
data sets, or n-way data partitioning, the basic function
of TDIDT is to discover the underlying rules within the
data presented. This requires a significant processing
operation to test the possible rule combinations. As a
result, many applications are confined to small data sets,
which raises questions as to how far the rules found can
be applied to the overall data. A significantly larger
data set reduces the influence of individual samples but
imposes unacceptable time penalties. ICEWARE® processes
larger data sets in shorter time frames, dramatically improving
both accuracy and operation. |
|
|
|
