Class Strength Prediction Method for Associative Classification

Test data prediction is about assigning the most suitable class for each test case during classification. In Associative Classification (AC) data mining, this step is considered crucial since the overall performance of the classifier is heavily dependent on the class assigned to each test case. This paper investigates the classification (prediction) step in AC in an attempt to come up with a novel generic prediction method that assures the best class assignment for each test case. The outcome is a new prediction method that takes into account all applicable rules ranking position in the classifier beside the class number of rules. Experimental results using different data sets from the University of California Irvine (UCI) repository and two common AC prediction methods reveal that the proposed method is more accurate for the majority of the data sets. Further, the proposed method can be plugged and used successfully by any AC algorithm.