Improving Performance of Ensemble Learners for Breast Cancer Detection Using Feature Engineering

Machine learning (ML) approaches include a variety of statistical and probabilistic methodologies that enable intelligent systems to be trained from repeated prior knowledge to find and recognize interesting patterns. Breast cancer (BC) is a form of tumour that grows in the tissues of the breast, and it is the most recurrent kind of disease across the world and one of the major reasons for fatality in women. Early identification of breast cancer may raise the chance of successful therapy and lower the mortality rate. In this study, the effectiveness of various ensemble approaches for the automatic prediction of breast cancer is compared and evaluated. The effectiveness of the learning process has been improved using Principal Component Analysis (PCA), a feature selection technique, by eliminating redundant and non-essential features. Diagnosis of Breast Cancer is achieved by utilizing the concept of Ensemble Learning (EL), an area of ML, including models like Gradient Boosting (GB), Adaptive Boosting (ADB), Extreme Gradient Boosting (XGB), and Random Forest (RF). The metrics that are utilized to analyze and evaluate the classifiers are ROC-AUC, Accuracy, Recall, Precision, and F1-score. The experimental results demonstrate that the Extreme Gradient Boosting is more accurate in predicting breast cancer, with an accuracy of 99.42%, compared to other ensemble learning algorithms.