A study of the robustness of association rules

摘要: Association rules discovery is one of the most important tasks in Knowledge Discovery Data Bases. The produced with APRIORI -like algorithms are then usually used for decision aiding expert and knowledge based systems and/or by a human end user. Unfortunately such may produce huge amounts thus steps association nowadays evaluation interpretation their interestingness. Objective measures provide numerical information on quality rule said "of quality" if its measure greater than user defined threshold. In this paper we propose new specificity objective interestingness measures: threshold sensitivity. By dealing problem intend to means measuring strength/robustness interest rule. We general framework allowing us determine number examples that can lose while remaining acceptable, panel classical transformation confidence. Keywords- rules, robustness