An early diagnosis of breast cancer (BC) is necessary for reducing the death rate among females across the globe. Since it takes a considerable amount of time to diagnose BC, it is required to develop an automated approach to the diagnosis of BC and, further, its classification into benign and malignant. Although several approaches based on machine learning (ML) models exist, still there is requirement of improvement accuracy and precision. Thus, a soft voting-based ensemble classifier is suggested in this paper to predict benign and malignant cancer by applying different classifiers. Owing to the strengths of classifiers, K-Nearest Neighbor (KNN), Gaussian Naïve Bayes (GNB), Logistic Regression (LR), and Support Vector Machine (SVM) are used in the ensemble classifier. The efficacy of the approach is determined by considering the widely used Wisconsin Diagnostic BC dataset (WDBC) in terms of performance parameters such as recall, precision, accuracy, and F1-Score. WDBC dataset is imbalanced, with more data being of benign cancer; the SMOTE approach is used for balancing this dataset before applying the classifier to avoid any bias in the results. The approach achieved recall and F1-Score of 99%, an accuracy of 98.59%, and a precision of 98%, which is better when compared to existing models. Further, the ten-fold cross-validation achieves an accuracy of 96.92%, thus demonstrating the robustness of the proposed work. Further, the ensembles can be applied to other types of approaches that are involved in the diagnosis of a specific disease and its severity.