Classification by essential emerging patterns in two phases

PNrule is a new two-phase framework for learning classifier models in data mining. The first phase detects the presence of the target class, while the second detects the absence of the target class. Emerging patterns (EPs) are itemsets whose supports change significantly from one data class to another, while essential emerging patterns (eEPs) are minimal ones among the EPs. It has been shown that the eEPs are useful and sufficient for building accurate classifiers. This work proposes a novel classification method, called CEEPTP, to combine the idea of two-phase induction and classification by EPs. The experiment study carried on 15 benchmark datasets from the UCI machine learning repository shows that CEEPTP performs comparably with other classification methods such as CBA, CMAR, C5.0, NB, and CAEP in terms of overall predictive accuracy.