Andreas Arning

Data Mining: Concepts and Implementation Issues

Over the last years the interest to extract “information” and “knowledge“ from data collected over time has grown tremendously. Es- pecially in the business area the need to use data from the past to drive the business successful in the future has created the demand to “mine“ data for patterns (i.e. relationships between the data of any form) that could guide future business actions as part of decision support systems. In contrast to verification based methods (like e.g. OLAP), data min- ing methods use discovery as underlying principle. This means, queries against the data are allowed to be less specific, which is why data mining methods have the potential to reveal surprises, unanticipated informa- tion, from the data. But since data mining methods benefit from the redundancy inside the data, for best results any implementation should be able to process large data, large both in the number of rows and in the number of columns.