Coronary artery disease (CAD) is a most common type of heart disease. CAD happen when a blood clot cuts off the heart’s blood supply, causing permanent heart damage. Diagnosis of CAD can be done using angiography, echocardiogram, electrocardiogram, which are complex methods. Therefore, studies are done to predict CAD using machine learning algorithms. This study proposes, feature selection by particle swarm optimization(PSO) and Bat algorithms, clustering using K-means and classification using Multinomial logistic regression (MLR) and support vector machine (SVM) algorithms. This technique is cross checked upon 14 attributes with 303 instances. A benchmark dataset from Cleveland heart disease data is used. The Bat-SVM model achieves the highest prediction accuracy of 97 %. The proposed model has an increased accuracy from the existing systems.