This paper deals with the topic of application of machine learning in the medical branch of oncology for the purpose of automated patient diagnosis. In oncology, special importance is attached to detecting the disease at an early stage and making the diagnosis as accurate as possible. In the theoretical part of the paper, problems in oncological diagnostics are described with special emphasis on breast cancer, and an overview of research and progress in the field achieved using machine learning is given. Basic concepts from the field of machine learning are introduced and the working principle of two learning algorithms, SGD and SMO, is described. In the empirical part of the work, the mentioned algorithms were applied to the breast cancer data set, which contains information on the types and properties of tumors. The data were pre-processed and analyzed using the Weka software tool, and finally the achieved results were presented. The results on the data used in this work show a satisfactory level of accuracy in determining the type of tumor.