Predicting Breast Cancer by Analysing Wisconsin Dataset Through Machine Learning Algorithms Using Orange Tool
Anamika Saroha, Aashish S Kamath, Monesh Kumar, T. R. Saravanan, N. Kanimozhi · 2025
Breast cancer is still a major global health issue, and early detection is important for better patient outcomes. This research discusses the use of machine learning algorithms for predicting breast cancer through the Orange data mining software. We utilized a comprehensive dataset containing various breast cancer features to develop and evaluate multiple machine learning models. The Orange tool facilitated data preprocessing, feature selection, model training, and performance evaluation. Our analysis compared the efficacy of several algorithms, in-cluding random forests, support vector machines, KNN, Gradient Boosting and neural networks. The study highlights the potential of orange tool for machine learning in enhancing breast cancer detection and diagnosis, potentially aiding healthcare profession-als in making more informed decisions. Furthermore, the user-friendly interface of the Orange tool showcases its utility in medical research and potential for wider adoption in clinical settings. For this paper, we consulted the Wisconsin dataset. The findings reveal that SVM models, particularly those utilizing ensemble methods like bagging and boosting with RBF kernels, significantly outperform traditional classifiers in both small and large datasets. In contrast, while Neural Networks demonstrated competitive accuracy, they often faced challenges related to overfitting and computational demands. Feature selection techniques were identified as crucial for enhancing model performance, allowing for more efficient training processes.