Prediction of breast cancer based on RF, SVM and PCA

Abstract

Breast cancer is often called the pink killer. The most recent statistics from the 2018 International Agency for Research on Cancer (IARC) survey show that the incidence rate of breast cancer is 24.2%, placing it first among women’s cancers globally, with 52.9% of cases occurring in developing nations. When cells in the breast start to grow abnormally, it can lead to the formation of tumours. These tumours can be detected through X-ray imaging or felt as lumps in the breast area, so early detection and treatment are the key to reducing the mortality rate of breast cancer. In this paper, machine learning techniques are employed to construct a breast cancer prediction model using the breast cancer dataset. Different algorithms were utilized in this paper, e.g., 3 machine learning algorithms - Support Vector Machines (SVM), Random Forest (RF), and Principal Component Analysis (PCA) - were utilized to build models for this task. The resulting models achieved impressive performance metrics, with accuracy rates of 94%, 97%, and 96% respectively. The author also finds that some features in the data are insignificant for prediction, and discarding them can lead to faster training time. However, more data is needed to achieve better performance in actual scenarios. These models can help doctors diagnose breast cancer more quickly and accurately.