A novel technique for breast cancer classification based on feature reduction and machine learning

Today, due to increased urbanization, industrial pollution and modern life style, cancer has become a challenge to humanity. Breast cancer is one of the most widespread, and yet fatal disease of modern time. Every year, a large number of people (mostly women) Affected and die from breast cancer. And it has the highest annual mortality rate between cancers among women. Breast cancer treatment requires accurate diagnosis on early stages, and since knowledge-based systems are useful for diagnosing diseases, we present a brand new system (T-RF) for diagnosing breast cancer. In this study, we begin with dimension reduction and reducing the number of attributes to several important features by t-Distributed Stochastic Neighbor Embedding (T-SNE), and then we use our chosen classifier which is Random Forest (RF) to classify our data. The data sets used in this study are Wisconsin Breast Cancer Data set (WBCD) and Wisconsin Diagnosis Breast Cancer (WDBC). And eventually, we acquire the required parameters and compare them with some of existing methods.

Keywords:

Breast cancer , Knowledge-based system , Dimension reduction.