MPGN - An Approach for Discovering Class Association Rules

Data mining is of great importance in the overall process of knowledge discovery. Different rule quality measures and rule ordering schemes can be applied in the process of rule selection. The mining of association rules is a typical data mining task that works in an unsupervised manner. A major advantage of association rules is that they are theoretically capable to reveal all interesting relationships in a database. Class-Association Rules (CAR) algorithms have a special place within the family of classification algorithms. This type of classifiers offers a number of advantages: efficiency of the training regardless of the training set; easy handling with high dimensionality; very fast classification; high accuracy; classification model easily comprehensible for humans. This work describes a new associative classifier - MPGN, that effectively builds and stores the set of rules in multilayer structures. The experiments performed proved the viability of the proposed approach, showing good performance of MPGN compared to other similar rules-based classifiers, especially in the case of large-volume data analysis with many classes and unevenly distributed.