A comparative assessment of machine learning algorithms with the Least Absolute Shrinkage and Selection Operator for breast cancer detection and prediction

Abstract

Breast cancer is the most common life-threatening cancer in women and one of the leading causes of death. Early diagnosis is one of the best defenses against the spread of breast cancer. Machine learning (ML) tools are now available for cancer detection and prediction. This study presents a comparative assessment of machine learning models for diagnosing breast cancer based on various classification schemes. Our classification methodology is based on well-organized data collection, preparation, transformation, and exploratory analysis (including correlation matrix, histogram, and data distribution). All characteristics are compared with the results of applying the Least Absolute Shrinkage and Selection Operator (LASSO) approach, which selects the most important attributes. Logistic Regression (LR), K-Nearest Neighbors (KNN), Extreme Gradient Boosting (XGB), Gradient Boosting (GB), Random Forest (RF), Multilayer Perceptron (MLP), and Support Vector Machine (SVM) algorithms have been applied in this study. We have achieved the maximum accuracy of 90.68% by RF compared with LASSO. Similarly, the recall in KNN was 98.80%, the precision in MLP was 92.50%, and the F1 score in RF was 94.60%.