Coronary artery disease (CAD) causes millions of deaths worldwide every year. The earliest possible diagnosis is quite important, as in any diseases, for heart diseases causing such a large amount of death. The diagnosis processes have been more successful thanks to the recent studies in medicine and the rapid improvement in computer sciences. In this study, the goal is to employ machine learning methods to facilitate rapid disease diagnosis without the need to observe negative outcomes. The dataset utilized in this study was obtained from an IEEE DataPort data repository. The dataset consists of two classes. Firstly, new features have been produced by using the features in the dataset. Then, datasets that consist of multiple features have been created by using feature selection algorithms. Three models, specifically Support Vector Machines (SVM), the k-Nearest Neighbor algorithm (kNN), and Decision Tree ensembles (EDT), were trained using custom datasets. A hybrid model has been created and the performances have been compared with the other models by using these models. The best performance has been obtained from SVM and its seven performance criteria in order of accuracy, sensitivity, specificity, F- measurement, Kappa and AUC are 97.82, 0.97, 0.99, 0.98, 0.96 and 0.98%. In summary, when evaluating the performance of the constructed models, it has been demonstrated that these recommended models could aid in the swift prediction of coronary artery disease in everyday life.