Finding Significant Frequency Changes in Large Databases

Abstract

Data mining can discover significant frequency changes between two large databases. When multiple data sources are used it is important sometimes to know the subtle differences between the databases. Such knowledge can help in making well informed decisions. This kind of mining is known as contrast mining. Pattern which is not in one dataset and present in another dataset is known as jumping emerging pattern. On the other hand, the pattern which has less frequency in first dataset and more frequency in other dataset is known as emerging pattern. In this paper, a general framework for building robust and accurate classifier is presented which guides in making EP based classifier. Afterwards, an algorithm is proposed for making a model that can help in classification of testing instances. We have built a prototype application that demonstrates the proof of concept. The empirical results are compared with other classifiers like NB, SVM and C4.5.