Amalgamation of SVM Based Classifiers for Prognosis of Breast Cancer Survivability

For last few years, researchers are increasingly employing machine learning methods in the domain of cancer prognosis. The main reason behind these efforts is to help oncologist to make accurate and less invasive decisions for the patient's treatment. Moreover, it would relieve many cancer patients from agonizingly complex surgical treatments and their colossal costs. In this paper, we have proposed an amalgamation method to form a composite classifier for predicting the survival chances of breast cancer patients. The composite classifier architecture takes classification results in the form of distance information of data samples from the hyper planes, accuracy values and a list of support vectors from individual SVMs to generate combined classification decision output. We show that this would help to achieve better classification results for breast cancer prognosis.