Designing of Augmented Breast Cancer Data using Enhanced Firefly Algorithm
P. Sri Silpa, A. Mallikarjuna Reddy, Ch. B. V. Durga, Vanga Navya Sri, U. Bhavani, Ch. Hema Harsha Priya, J. Mounika
Abstract
Breast cancer is the world’s most widespread disease. Specific hereditary gene mutations considerably increase breast cancer risk. Treatment for breast cancer can be quite beneficial, with survival rates of Ninety percent or greater, specifically when the disease is detected early. In the system to identify cancer in its early stages, computational techniques are used to create augmented data based on its importance. Based on the specified features in the dataset, the system employs multiple machine learning techniques to predict whether the cancer is mild or severe. The proposed method uses these three datasets (Wisconsin, Prognostic, and Coimbra), provided in the Kaggle and UCI data repositories. optimization for feature selection performs on these three datasets. This model uses a new metaheuristic algorithm called the Enhanced Firefly algorithm, which is inspired more by illuminating behavior of fireflies. With this algorithm used in the system the Wisconsin dataset reduces from 32 to 6 features, the Prognostic dataset reduces from 32 to 9, and the Coimbra dataset reduces from 10 to 4. This Enhanced firefly algorithm uses accuracy and reduction of features as fitness functions. The major goal of this model is to design a generalized dataset that can work to predict the optimal accuracy using any stacking classifiers.