Imbalanced Data Using with-in Class Majority Under Sampling Approach

The existing data can be further processed by using data mining techniques for novel discovery of knowledge. Classification is one of the techniques used for knowledge discovery. One of the most popular models used for knowledge representation is decision trees due to its easy of interpretation. Class imbalance data sources are of the critical and most popular data sources for their real time availability. The existing classification techniques are bottle necked in the scenario for skewed distributed data. To address his short coming, we have proposed a novel algorithm With In class Majority Under Sampling (WIMUS) for improved learning. The results generated are encouraging to show that our proposed approach is effective on class imbalance data sources.