Ensemble of XGBoost Classifiers Based on LDA Dimensionality Reduction for Predicting Breast Cancer
Mai Nhu Uyen Le, Jianlin Zhou, Dinh Phu Cuong Le, Dong Wang · Universal Journal of Public Health · 2024
As reported by the World Health Organization, breast cancer is recognized as the most popular disease in women. Thus, the need for early and accurate detection of this cancer for effective treatment is highly demanded. In this paper, a novel machine learning-based method is proposed to improve the success of breast cancer prediction. To be specific, Extreme Gradient Boosting (XGBoost), which is an efficient machine learning algorithm to deal with large datasets, is applied with the help of the Linear Discriminant Analysis (LDA) algorithm, which is often used for dimensionality reduction by fusing the original multidimensional data features, to create the cancer predictive model. From the experimental results, with the LDA, it is shown that the XGBoost classifier can help to improve the classification accuracy by 2.7 % compared to the classifier without using LDA. Moreover, when compared to other machine learning methods, the proposed method also shows a better classification result with the root mean squared error of 0.115, which means that its error is at least 2.6 % lower than others. The proposed method aims to support doctors in enhancing clinical application as well as improving medical quality, especially when detecting the very first moment of breast cancer.