Analysis of Risk Factors for Cervical Cancer Based on Machine Learning Methods

Abstract

Cervical cancer is one of the illness which is threatening women's health all over the world and it is hard to observe any sign in the early stage. Three methods have been introduced in this paper to analyze the dataset of cervical cancer, including SVM (Support Vector Machine), XGBoost (eXtreme Gradient Boosting) and Random Forest. The dataset contains 32 risk factors and four target variables: Hinselmann, Schiller, Cytology, and Biopsy. And the diagnostic results of these four target variables were classified by the three methods that mentioned above. Finally, the top five risk factors which affect the diagnosis most were found, and the classification results showed that XGBoost and Random Forest perform better than SVM.