A novel approach to handle class imbalance : A Survey
Ms. Monica. Ochani,  Dr.S.D. Sawarkar,  Mrs. Swati Narwane
Machine learning is a study of algorithms that a system uses to effectively perform a specific task. It depends on the patterns and inference instead of any instructions. In machine learning, majorly there is some level of class imbalance issue in real-world classification. This problem arises when each class does not make up an equal division of a data-set. It is essential to properly alter the metrics and methods to balance the data set goals. This means that many learning algorithms of machine learning have low predictive accuracy for the not often occurring class. In this paper, we shall discuss this problem and look in to different approaches used to solve the class imbalanced issue. This paper discusses the survey of different approaches done to improve the class imbalance issue in the data sets by learning about the data level approaches and the algorithm approaches. We have discussed the oversampling and undersampling methods to overcome the data imbalance problem.
Keywords- Class imbalance, data mining, machine learning, imbalance data, applications, classification, approach, algorithm, sampling.
Cite this Article
Ms. Monica. Ochani,  Dr.S.D. Sawarkar,  Mrs. Swati Narwane,   "A novel approach to handle class imbalance : A Survey"
, International Journal of Engineering Development and Research (IJEDR), ISSN:2321-9939, Volume.7, Issue 2, pp.419-422, June 2019, Available at :http://www.ijedr.org/papers/IJEDR1902079.pdf