Selecting the Dataset for Classification Using Predictive Apriori and Diversity Measures

The main task of Association rule mining is to find correlations among the set of data items present in the database. Rule interestingness is mainly measured by means of support and confidence. There exists various other measures for depicting the rule interestingness such as Lift, Conviction, Drift etc. Apart from these, there also exists diversity measures which are applied on Summaries. Much little work was done on association rule mining using diversity measures. This article suggests the use of predictive apriori approach for selecting the best dataset based on the application of diversity measures on the association rules generated. The experimental results are encouraging.