Megaputer Announces a Final Release of PolyAnalyst 4.0

Megaputer announces the final release of PolyAnalyst 4.0, the newest version of the leading data mining solution. An evaluation copy of PolyAnalyst 4.0 can be downloaded from http://www.megaputer.com.

Version 4.0 represents a major upgrade of the system. PolyAnalyst now utilizes Distributed Component Object Model (DCOM) technology, features ten unique machine learning algorithms, and furnishes data manipulation, visualization, and scoring capabilities. In addition to clustering, predicting, dependency detecting, and yes/no classifying, PolyAnalyst 4.0 solves tasks of explicit modeling, detection of association rules in transactional data, and classification to multiple categories. An open DCOM architecture makes PolyAnalyst 4.0 easily extendable, upgradable, and customizable. It provides the user with an option to purchase only the necessary machine learning algorithms as individual modules and utilize these modules as an integral part of their data storage and management system. A DCOM-based PolyAnalyst Knowledge Server can support several client stations on a local network.

New features of PolyAnalyst 4.0 include:

• Interactive 3D graphics providing a better intuitive feeling for the data prior to running machine learning analysis.
• Unique Market Basket Analysis algorithm for processing transactional data. Groups of products sold together well and the corresponding directed association rules are identified - an order of magnitude faster than by traditional algorithms.
• Memory Based Reasoning algorithm based on a combination of Nearest Neighbor and Genetic Algorithms. The new method is used efficiently for classification into multiple categories, as well as prediction of numerical variables.
• Implementation of DCOM architecture. Individual PolyAnalyst algorithms can be integrated in the form of COM modules in external decision support or data management applications.
• Support for OLE DB for Data Mining, a coming standard from Microsoft, dramatically simplifying and unifying communications between RDBMS and Data Mining applications.
• Support for the analysis of large data sets. The maximum volume of data accepted by the system has been significantly increased and a new mechanisms for dealing with large data sets have been implemented.
• Redesigned user interface. The project contents are organized in a Windows-standard tree-like style, while preserving all the best features of the traditional PolyAnalyst interface.
• Dynamic HTML reports for exploration engines. The new interactive reports can be customized, saved, printed, or copied in a standard portable format, and exchanged with external applications
• Mouse-driven Rule Assistant for simple creation of user defined transformation rules.
• Lift and Gain Charts for interpreting the results obtained by machine learning algorithms are especially valuable for direct marketing tasks. These features are very useful for measuring the extra profit reaped by a marketer making decisions based on the knowledge discovered by PolyAnalyst.
• Snake Chart for convenient visual comparison of different data sets.
• Chart Designer for the development of custom charts and advanced data visualization capabilities
• PA Scheduler enabling batch process data mining in PolyAnalyst. The user can record a sequence of actions and schedule the created script to be run by PolyAnalyst at a specified time and on the specified data sets, applying the model to the specified data.
• Sampling capabilities, allowing random selection of records from a data set.
• Direct data exchange with Oracle Express and IBM Visual Warehouse.

"By offering these new features, Megaputer has clearly positioned PolyAnalyst 4.0 as the number one modern business solution for data analysis," says Sergei Arseniev, CEO of Megaputer Intelligence. "Now PolyAnalyst combines the broadest selection of versatile machine learning algorithms with the convenience and flexibility of DCOM architecture. The new cutting-edge technology helped Megaputer significantly increase its DM market share during the last year."