Research on approach for classification of Within imbalanced data sets

Most of the existing methods for unbalanced data classification only consider about the situation of imbalance between classes but don't consider about the situation within the class, thus affect the final classification results. In order to eliminate the imbalance within the class, put forward the cluster algorithms based on DBSACN algorithm to process the imbalance problem within the class. Through the determination of adaptive the ԑ and MinPts of DBSCAN algorithm, then form clusters and resampling within them. Then resolve the data fragmentation and density problems of the imbalance within class. Use UCI data for testing, and then compare the %ACC F-Measure and AUC with other algorithms to prove the effectiveness of the algorithm.