Introduction: The reports obtained after pathological examinations on breast cancer have been digitized by sophisticated machines and stored as Electronic Health Records (EHRs). These EHRs contribute to Computer-Aided Diagnosis and for better clinical decision support. Objectives: Understanding of various machine learning techniques used for the classification of standard and real-time breast cancer datasets. Reviewing of classification of various types of dataset like images, text, numerical values etc., on breast cancer. Methods: In this paper, a rigorous literature survey has been made on the classification of the dataset on breast cancer using various machine learning methods on standard datasets like Breast Cancer Dataset, Wisconsin Breast Cancer Dataset, Wis consin Breast Cancer Diagnostic Dataset, Wisconsin Breast Cancer Prognostic Dataset and Surveillance Epidemiology and End Results Dataset etc. In the literature, it has been observed that some of the authors have worked on the classification of datasets that are collected from different hospitals. Images of the breast have been analyzed by looking at the property of luminance, colour and shape variation, texture, reaction to biomarkers and many other factors. For understanding proliferation in breast cancer, various scoring systems are used. They include Bloom-Richardson Score, Masood Score, Modified Masood Score, Robinson__ampersandsignrsquo;s Score and many others. The EHRs containing the records in text form on breast cancer have been interpreted using Natural Language Processing approaches like text segmentation, named entity recognition and part of speech tagging etc., and classified using machine learning approaches. Results: Classification of breast cancer has been made on different types of datasets using machine learning methods and the range of accuracy obtained is between 75.60% and 99.86%. Conclusion: Most of the existing classifiers are binary classifiers to classify breast cancer datasets into benign and malignant classes. However, it is necessary to design multiclass classifiers for building a precise clinical decision support system and to provide targeted therapy for cancerous patients using cost-effective diagnostic methods.