A Review on Handling Imbalanced Data

Abstract

Computational synthesize of the metabolic pathway is take low cost while comparing with the direct trial and error laboratory process. In real world data, more or less all datasets having a skewed distribution of classes. The skewed and the number of instances for certain classes much higher than other classes, this problem is known as the class imbalance problem. Practically this class imbalance problem reduces the classification accuracy because it predicts the minority class instances inaccurately. Class imbalance is an issue encountered by data mining practitioners in a wide variety of fields. The classification of imbalanced data is a new problem that rises in the machine learning framework and it is the major problem raised for the researches and the use of sampling techniques to improve classification performance has received significant attention in related works. In this article the necessity of balancing an imbalanced data is elaborated and the methods proposed by the various authors for to balance the imbalanced data and the evaluation metrics to assess the accuracy and predictive rate of the classification algorithms also have been discussed