Hybrid Machine Learning Model for Breast Cancer Classification with K-Means Clustering Feature Selection Techniques

Breast cancer is one of the major causes of fatalities in women across the globe, necessitating early diagnosis and detection for successful treatment. Scientists have pinpointed numerous risk factors for breast cancer such as obesity and sex hormones. Machine learning prediction systems have employed diverse data types, like genetic profiles, clinical data, and X-ray images, to foresee the risk of breast cancer. To eliminate any unnecessary or superfluous attributes from high-dimensional datasets, algorithms for feature selection have been applied. This research proposes a composite machine learning model integrating five models: Gradient Boosting (GB), Random Forest (RF), Logistic Regression with Lasso Regularization, Logistic Regression with Ridge Regression, and Support Vector Machine (SVM). Feature selection methodologies employed are based on the K-Means clustering oriented feature importance measure. The Coimbra breast cancer dataset, composed of 116 data points and 10 characteristics, is used for training and assessing the models on their predictive abilities. According to the study's outcomes, the Support Vector Machine model demonstrates the greatest effectiveness for the dataset when used with the proposed feature selection, whereas Gradient Boosting and Random Forest are efficacious models without feature selection.