Addressing imbalance problem in the class - A survey
Tamil Nadu · 2014
In recent years, major challenges have been evolved for the classification of imbalance data. Classification of imbalance data is very tedious based on the nature and size of the data. Class imbalance problems, mainly occur when the samples in one class having more sample than the other class. The class which has large samples is known as majority classes and the class which has the least number of samples is known as minority classes. Mostly many algorithms focus mainly on majority classes by eliminating by minority classes; this does not allow the algorithm to obtain good performance. We can divide the classification of imbalance dataset into three categories. Each of these approaches has its own pros and cons. In this work, the systematic study for each approach has been done. Keyword: Data imbalance, Classification, Sampling Methods, Ensemble Methods.