Abstract: Data mining also called knowledge discovery is essential for solving problems in specific domain. Medical diagnosis is a time consuming and difficult task performed by experts. For improving medical diagnosis using data mining techniques, data mining techniques need to be applied on relevant features. Over fitting can be reduced and performance of mining can be enhanced if more relevant features are selected. This paper reviews the various feature selection techniques and their application in different domains.

Keywords: Relief Attribute Evaluation, Symmetric uncertainty attribute, Multilayer Perceptron