OPTIMIZED BREAST CANCER PREDICTION USING MACHINE LEARNING ALGORITHMS

ABSTRACT Breast cancer remains one of the leading causes of mortality among women globally, making early and accurate diagnosis critical for effective treatment. This study presents an optimized machine learning approach for predicting breast cancer malignancy using the Wisconsin Breast Cancer Dataset (WBCD). The methodology involves a systematic pipeline comprising data preprocessing, feature selection, model development, hyperparameter optimization, and model evaluation. Four classification algorithms; Logistic Regression, Decision Tree, Random Forest, and Support Vector Machine (SVM), were implemented to classify tumors as benign or malignant. Data preprocessing techniques such as normalization, missing value imputation, and feature encoding were applied to prepare the dataset. Feature selection methods, including filter, wrapper, and embedded techniques, were used to enhance model performance and reduce overfitting. Optimization strategies like grid search and 10-fold cross-validation further improved the models’ generalizability. The results demonstrated that Random Forest and SVM significantly outperformed the other models in terms of accuracy, precision, recall, and F1-score, achieving predictive accuracies above 96%. In contrast, Logistic Regression showed poor performance, underlining the importance of model complexity in medical prediction tasks. Evaluation using ROC and precision-recall curves confirmed the robustness of the best-performing models.