Comparison on Three Supervised Learning Algorithms for Breast Cancer Classification

Breast cancer is a widespread and potentially deadly illness that affects many women worldwide. The uncontrolled growth of cells in breast tissue results in the formation of tumors, categorized as malignant or benign. Malignant tumors pose the greatest threat due to their potential to result in fatality. Early detection becomes crucial in preventing adverse outcomes, ensuring swift access to suitable treatment, and mitigating the progression of tumors. Traditionally, breast cancer diagnoses relied on medical expertise, introducing the risk of human error. Hence, this study addresses this challenge by employing supervised machine-learning models using a Support Vector Machine (SVM), K-nearest neighbors (KNN), and Random Forest (RF), to classify breast cancer as benign or malignant. These algorithms are chosen due to their proficiency in handling classification tasks. This study aims to implement 10-fold Cross-validation to the three models chosen to ensure model robustness applied to Breast Cancer Wisconsin (Diagnostic) dataset. Accuracy score is chosen as the performance measures for the models. Based on the results, RF with 500 and 1000 estimators outperformed SVM and KNN with an accuracy of 96.31%, compared to SVM (95.61%) and KNN (93.15%). To conclude, this study’s outcomes have the potential to significantly contribute to the development of an automated diagnostic system for early breast cancer detection.