Equating Interestingness of Causal Rules via Graded Response Theory
Shinichi Hamano · 2006
Multi-database mining has attracted a lot of attention because it is an important research topic for large companies that have many branches to generate powerful insights that lead to benefits. However it is difficult for existing algorithm to generate both global and local patterns and compare interestingness of patterns because there is no unified measures in data mining area. This paper proposes a method of equating interestingness of patterns for extracting and comparing both global and local patterns via unified measure latent trait based on graded response theory.