Breast Cancer Prediction: Importance of Feature Selection

Abstract

In today’s world, breast cancer is one of the most widespread causes of death in women. According to an estimation, approximately 40,920 women would die in 2018 just because of breast cancer, which is a highly alarming number. Such alarming numbers could be reduced if the cancer is diagnosed at an early stage. With the advent of technology, making such predictions has become an easier task. Machine learning is one of the latest trends, which enables to make predictions related to diseases based on physical or behavioral characteristics. In this paper, we use various machine learning algorithms like decision trees, k-nearest neighbor (KNN), logistic regression, neural networks (NNs), naïve Bayes, random forest, and support vector machine (SVM). The outcome is then compared based on the precision, recall, and F1 score. Furthermore, we identify the least important features in the dataset, implement all these algorithms again after removing those features, and then compare the outcomes for the two implementation stages in order to understand the importance of feature selection in breast cancer prediction.