Breast Cancer Prediction Based on Machine Learning

Abstract

Breast cancer is a significant health concern, necessitating accurate prediction models for early detection and improved patient outcomes. This study presents a comparative analysis of three machine learning models, namely, Logistic Regression, Decision Tree, and Random Forest, for breast cancer prediction using the Wisconsin breast cancer diagnostic dataset. The dataset comprises features computed from fine needle aspirate images of breast masses, with 357 benign and 212 malignant cases. The research findings highlight that the Random Forest model, leveraging the top 5 predictors—“concave points_mean”, “area_mean”, “radius_mean”, “perimeter_mean”, and “concavity_mean”, achieves the highest predictive accuracy of approximately 95% and a cross-validation score of approximately 93% for the test dataset. These results demonstrate the potential of machine learning approaches in breast cancer prediction, underscoring their importance in aiding early detection and diagnosis.