Applying Dependency of Attributes for Business Data Mining in Information Systems: A Theoretical Framework

Abstract

This paper presents the applications of dependency of attributes in information systems for data mining from business datasets. Firstly, we present the theoretical framework for data clustering on small business dataset. It is based on a construction of a hierarchical rough set approximation in an information system for data splitting. The hierarchy is defined by the notion of a nested sequence of indiscernibility relations that can be defined from the dependency of attributes. Secondly, an application of such hierarchy for mining maximal association from a business transactional data is presented. It is shown that the dependency provides clear and provable theoretical approach for data clustering and maximal association rules mining.