Review of Imbalanced Data Classification and Approaches Relating to Real-Time Applications

Abstract

In today's era, multifarious data mining applications deal with leading challenges of handling imbalanced data classification and its impact on performance metrics. There is the presence of skewed data distribution in an ample range of existent time applications which engrossed the attention of researchers. Fraud detection in finance, disease diagnosis in medical applications, oil spill detection, pilfering in electricity, anomaly detection and intrusion detection in security, and other real-time applications constitute uneven data distribution. Data imbalance affects classification performance metrics and upturns the error rate. These leading challenges prompted researchers to investigate imbalanced data applications and related machine learning approaches. The intent of this research work is to review a wide variety of imbalanced data applications of skewed data distribution as binary class data unevenness and multiclass data disproportion, the problem encounters, the variety of approaches to resolve the data imbalance, and possible open research areas.