Selecting Feature Subset via Constraint Association Rules

摘要

In this paper, a novel feature selection algorithm FEAST is proposed based on association rule mining. The proposed algorithm first mines association rules from a data, set; then, it identifies the relevant and interactive feature values with the constraint association rules whose consequent is the target concept, and detects the redundant feature values with constraint association rules whose consequent and antecedent are both single feature value. After that, it eliminates the redundant feature values, and obtains the feature subset by mapping the relevant feature values to corresponding features. The efficiency and effectiveness of FEAST are tested upon both synthetic and real world data sets, and the classification results of the three different types of classifiers (including Naive Bayes, C4.5 and PART) with the other four representative feature subset selection algorithms (including CFS, FCBF, INTERACT and associative-based FSBAR) were compared. The results on synthetic data sets show that FEAST can effectively identify irrelevant and redundant features while reserving interactive ones. The results on the real world data sets show that FEAST outperformed other feature subset selection algorithms in terms of average classification accuracy and Win/Draw/Loss record.