Nowadays the use of computer technology in the field of medical diagnosis and prediction of disease has increased. In these fields the computers are used with intelligence such as fuzzy logic, artificial neural network and genetic algorithms. Many techniques of data mining are useful in the field of medicine and many algorithms have been developed. The main objective of this work is to find out the important attributes which are highly important for accuracy of the classifier and reduce the dimensionality of dataset for classification of disease dataset. The other objective of this work is to classify the dataset in cost effective manner. As many tests are redundant and also are highly expensive. We have used various approaches for feature selection as using Brute force approach and correlation based approach. We have also proved that accuracy of classifiers are improved using feature selection.