Software

FQS Poland, Duchsoft and Fujitsu Kyushu System Engineering announce the worldwide launch of a superior new software package, GhostMiner for data mining and visualization (www.fqspl.com.pl/ghostminer)

GhostMiner is advanced, analytical data mining software from Fujitsu that enables more efficient use of, both corporate and scientific, databases through the building of data models. The software also comes as the tool for visualization and analysis of multidimensional data.

GhostMiner is based on most advanced artificial intelligence solutions. The software's primary tasks include learning, memorizing, and utilizing the accumulated knowledge hidden in vast databases. This knowledge may then be used to address organization-specific operational issues.

The GhostMiner package includes two tools: Developer and Analyzer.

GhostMiner Developer is the tool for data-model designers and developers who, using databases, can train, test, run, and select useful models. The use of this program requires good knowledge of statistical analysis and some knowledge of methods of computational intelligence.

GhostMiner Analyzer is for end-users who are not experts in statistics or data mining but simply want to learn about their data using data models created with GhostMiner Developer.

GhostMiner is characterized by the following features:

• Intuitive interface for managing projects of arbitrary complexity, in the form of project windows and project trees. Projects store all results of experiments performed, automatically adding models developed during cross-validation.
• A wealth of statistical information and charting options to view the data, facilitating quick detection of outliers (errors or non-typical cases). Data may be viewed in original or standardized/normalized formats, filtered and ordered in a variety of ways. features, and 2-dimensional projections for viewing of data points for any combination of features.
• Several complementary adaptive models, including neural networks, neurofuzzy system, decision tree, and the k-nearest neighbor algorithm.
• Additional methods to enhance data models by creating committees composed of arbitrary models or subcommittees of models, and k-classifiers composed of single models or committees, one for each class.
• Variety of methods for testing accuracy and efficiency of trained models (cross-validation, Xtest) along with the confusion matrix summarizing the model quality.
• Interactive multidimensional data visualization tool (IMDs). GhostMiner is capable of extracting useful information that otherwise remains invisible and more:
• Discovering knowledge in the form of decision trees or logical rules, both crisp or fuzzy.
• Finding missing values in the data and complete patterns with missing parts.
• Automatic discretization of input variables in the case of fuzzy rules.
• Selecting variables important for a given decision or assigning weights to input variables, according to their importance.
• Making discrete continuous inputs providing intervals useful for symbolic description of data.
• Visualizing important relationships in data sets, showing data in relation to other known cases stored in the database.
• Calculating probabilities of assigning new cases to existing categories.
• Finding explanations for such assignment by using models that provide logical rules.
• Showing similar cases using interactive visualization, reminding of other possibilities that may often be overlooked.
• Finding cases that are most similar in respect to variables that are important for a given category.
• Using projects based on committees of models to evaluate ambiguous cases.

About FQS Poland Sp. z o.o.:

FQS Poland is a Fujitsu Group software research and development company. We specialize in research, development, sales and support of scientific software. FQS Poland focuses on research and development of new software products in the fields of computer chemistry, artificial intelligence in technical sciences, as well as computer aided design and computer aided manufacturing.

About Duchsoft Duchsoft is a group of experienced PhD scientists supervised by Professor Wlodzislaw Duch - a world recognized expert in computational intelligence area, a member of the editorial board of the IEEE Transaction on Neural Networks journal, frequently providing tutorials on data mining at the most important conferences and congresses around the world.

Contacts: For more information please contact: E-mail: gm@fqspl.com.pl URL: http://www.fqspl.com.pl