Breast Cancer Classification Using Optimizer-based Feature Selection: A Metaheuristic Approach
Anishkaa Balasubramaniam, Lackesh Shanmugasundaram, T K Divya, Hemprasad Yashwant Patil · 2023
Breast cancer, in particular, has been acknowledged as being among the deadliest conditions among females, accounting for one-third of all female deaths. Cancer cells spread throughout the body by entering the circulatory or lymphatic systems. As a result, identifying and analyzing ways to aid in understanding breast cancer and its impact on women becomes critical. This work aims to differentiate and categorize two main sorts of cellular growths: benign and malignant, using information gathered by the Wisconsin Breast Cancer Diagnostic Collection. Many people refer to benign tumors as “non-cancerous,” while malignant tumors are “cancerous.” Six nature-inspired feature selection algorithms are employed for this purpose: Zebra Optimization Algorithm, Pelican Optimization Algorithm, Tasmanian Devil Optimization Algorithm, Northern Goshawk Optimization Algorithm, Dwarf Mongoose Optimization Algorithm, and Walrus Optimization Algorithm. The chosen algorithms are contrasted with outcome measures often used in evaluating machine learning models including accuracy, scores such as F1, precision, and recall.