 |
Coral Uniqueness
Componentised architecture: Hybridisation in mind
Coral consists of a set of optimised mathematical
components aimed at data transformation through
to computationally complex data mining modules.
The architecture of the components has taken into
account the fact that hybrid models are often required
to solve difficult data mining problems.
Hence, while Coral provides a base set of data mining
algorithm components, the object-oriented architecture
of Coral allows for these base algorithms to be
integrated to quickly produce the data mining solution
most applicable to the problem at hand. The GENNA
algorithm and its application to Property Valuation
is an example of such hybridisation. The base technologies
used by GENNA include: Nearest Neighbour, Genetic
Algorithm, Clustering and Survival Analysis.
Standards Compliance: Ease of Integration: Speed
of Delivery: Lower Costs
Corals data mining components have been developed
with integration in mind. The Coral development
team are participants on standards committees including
the Predictive Modelling Markup Language (PMML)
and the Java Data Mining API (JDMAPI). All data
mining components that generate models that can
be represented in PMML provide import/export facility
from/to PMML.
Those parts of the models that are currently not
supported by PMML are represented in XML. CORAL
will also be an early adopter of the JDMAPI, which
is currently under development as part of the Java
Community Process, when it is released. When these
developments are technically interesting, the key
reason for CORAL actively supporting these standards
are to provide its users with the business benefits
of speed of solution delivery and the ease and resulting
lower costs of integration within their IT infrastructure.
Parallelisability
As data volumes continue to swell the speed with
which data mining algorithms can analyse the data
will be critical. With this in mind the data mining
components within CORAL have been design with parallel
execution and distributed computing in mind. This
fact, coupled with the standards compliance of these
components, makes the development and delivery of
data mining solutions tailored to the customers
needs possible and efficient.
back
to top
back to products |
Coral
Summary
