Prediction and correct voting is critical task in imbalance data multi-class classification. Accuracy and performance of multi-class depends on voting and prediction of new class data. Assigning of new class of imbalance data generate confusion and decrease the accuracy and performance of classifier. Various authors and research modified the multiclass classification approach such as one against one and one against all. In both method OAO and OAA create a unclassified region for data and decrease the performance of classifier such as support vector machine. Some other method such as decision tree classifier, nearest neighbor and probability based classifier also suffered from imbalance data classification. In this paper we discuss various method and approach for multi-class classification for imbalance data.