Adaptive Synthetic Oversampling Algorithm for Handling Class Imbalance in Multi-Class Data Stream Classification

Concept drift and class imbalanced data are major challenging processes involved in modern streaming classification. Particularly, when integrated with difficult factors like the existence of noise, overlapping distribution, concept drift, imbalance can considerably affect classifier results. In addition, various challenges performance existing oversampling schemes such as SMOTE its derivatives. Regardless that, several models concentrate on binary classification problems, whereas complex multi-class counterparts yet to be explored. With this motivation, study develops an Adaptive Synthetic Oversampling Algorithm (ASYNO) based Multiclass Streaming Data Classification (ASYNO-MCSDC) model Class Imbalance Handling Drift. The presented ASYNO-MCSDC method initially performs different stages preprocessing label encoding, normalization, splitting. Besides, technique is applied for handling problems. Also, online bagging ensemble employed process which Hoeffding Tree (HT) was utilized base number estimators used set 10. For experimentation, two types learning used, one batch other incremental learning. experimental validation tested using datasets namely stationary stream dynamic stream. results pointed out that has accomplished promising over models.