Discovering Frequent Gradual Itemsets with Imprecise Data.

摘要: The gradual patterns that model the complex co-variations of attributes form "The more/less X, Y" play a crucial role in many real world applications where amount numerical data to manage is important, this biological data. Recently, these types have caught attention mining community, several methods been defined automatically extract and from different models. However, are often faced problem managing quantity mined patterns, practical applications, calculation all can prove be intractable for user-defined frequency threshold lack focus leads generating huge collections patterns. Moreover another with traditional approaches concept gradualness just as an increase or decrease. Indeed, considered soon values attribute on both objects different. As result, numerous quantities extracted by algorithms presented user although their only noise effect To address issue, paper suggests introduce thresholds which consider In contrast literature approaches, proposed approach takes into account distribution values, well user's preferences makes it possible certain databases fail due too large search space. Moreover, results experimental evaluation show algorithm scalable, efficient, eliminate do not verify specific requirements small set user.