First, the model generation was further automatised and simplified by means of automatic parameter tuning. An optimised feature configuration supports the faster model generation of the DISCOVERER especially for extremely large data and feature volumes.
Furthermore, a new database interface based on OLE DB was implemented, which extends the range of the directly accessible databases. Additionally, Version 5.0 provides an own binary storage format – the Prudsys Data File – which facilitates the storage of extensive meta information of the data.
The DISCOVERER is world’s most powerful classification tool for large data volumes. It applies Nonlinear Decision Trees as unified framework to combine different uni- and multivariate classification methods consistently. The model export via PMML and the real-time application by the PMML Applicator allow the seamless integration of DISCOVERER classification models into end-user applications. Since Version 5.0, the PMML Applicator is included into the DISCOVERER distribution.
A 30-day trial version of the DISCOVERER is available at
http://www.prudsys.de/...