BREAST CANCER PREDICTION VIA GRID SEARCH HYPERPARAMETER OPTIMIZATION

Abstract

Breast Cancer is a serious health issue worldwide, and early detection is crucial in preventing  deaths. Machine learning can help identify tumors efficiently, and this paper introduces the Grid  Search Hyperparameter Optimization (GSHPO) method to optimize the parameters of six existing  models, including Logistic Regression, Support Vector Machine, K-Nearest Neighbor, Naïve  Bayes, Decision Tree, and Random Forest. The best parameters were applied to predict outcomes  in six datasets, including OWBCD, WDBC, Coimbra, BRCA, Haberman, and SEER. The results  show that tuning the hyperparameters of models has a significant positive impact on prediction  accuracy.