Cardiovascular diseases (CVD) are a life-threatening group of diseases that contribute more to morbidity and mortality rates globally and have become very common recently. Unhealthy lifestyle, stress, cholesterol, and blood pressure are the factors affecting CVD. Machine learning (ML) approaches have been used to detect CVDs. The feature extraction process involves Synthetic Minority Oversampling Technique (SMOTE) and the mean substitution method. It uses a Support Vector Machine (SVM) algorithm for better results. Many ML methods, such as logistic regression, SVM, random forest, gradient boosting, decision trees help to detect CVD. To examine the proposed model, the authors experimented with the data set in individual classifiers, and then, concluded that the accuracy obtained from ensemble learning with hard voting resulted in 93% accuracy and was better than soft voting. Further, L1 Regularization was used to reduce the overfitting and bias.