Utilizing Machine Learning to Predict the Malignancy of a Breast Tumor

Abstract

Machine learning is creating computational models from data that predict outcomes and facilitate decisions. Using past data, machine learning can help predict the label of whether a current patient will more or less likely have breast cancer given only the features. This paper describes the process of creating a model to test if a patient has breast cancer using a dataset with data on the radius, texture, compactness, etc., of 569 breast tumors. One hot encoding, binary classification, validation and training sets, neural networks, binary cross-entropy error, accuracy, and the testing of epochs and different combinations of nodes, hidden layers, and activation functions will be included in this paper.