Author(s): Wakaa Ali Hadba and Hadeel Imad Naser

Abstract: In recent times, breast cancer has surpassed all other types of cancer to become the most widespread and primary cause of mortality among women globally. This study aims to forecast the benign or malignant nature of a breast tumor by employing various machine learning methods, including Support Vector Machines (SVM), Lasso Logistic Regression (LLR), Artificial Neural Networks (ANN), and Logistic Regression (LR). The findings of this research could potentially assist oncologists in accurately diagnosing the specific type of breast tumor. Various performance metrics were employed to assess the efficacy of training and validated models, including Accuracy or Classification Error, Sensitivity, Specificity, and Receiver Operating Characteristic (ROC). The models were constructed and evaluated to determine the optimal performing model. A dataset separate from the one used for model development was employed to validate each model. The data analysis results indicate that the Lasso Logistic Regression (LLR) model outperforms other models in classifying breast cancers. It exhibited superior performance in terms of accuracy, classification error, sensitivity, specificity, and ROC. Furthermore, it mitigates the issues of multicollinearity and high-dimensionality.

Pages: 83-89 | Views: 98 | Downloads: 24

Download Full Article: Click Here