Data mining is a way for extracting the valuable knowledge patterns from a huge amount of data. Various data mining tools and techniques are used in the medical world for predicting the diseases. Heart disease is one of the common diseases nowadays. This paper presents different feature selection attribute Evaluator models Working with all attributes in data is not always useful so for this study we filtered selected attributes that will give maximum accuracy. This study compares various classification techniques for predicting heart disease with a different set of attributes selected using evaluators. The dataset used for the study is Cleveland heart dataset it has 14 attributes and 303 instances. For achieving the results, the different classification algorithms are applied on selected attributes taken after feature selection. The classifiers used in the study are J48, SMO, Multilevel perception, Bagging Naive Bayes Random Tree. Then, a comparative study is done on the basis of accuracy achieved from the different classifiers.