Abstract:
Breast cancer is currently the most common cancer globally and exceedingly threatening pointedly amidst women. Due to the complexity of breast tissues, accurate discerning and categorization of breast cancer is a crucial medical endeavor. “Machine Learning (ML)” approaches are bringing success in a mixed bag of spheres. Discreetly among the health sector, because of their range to automatically cutting attributes. The principal goal in this research is to identify and classify breast cancer and its stages by applying “Machine Learning”. In this study we used variety of “Machine Learning” approaches specifically Logistic Regression (LR), Decision Tree (DT) Classifier, Random Forest classifier, Support vector Classifier (SVC), K-Nearest Neighbor (KNN) Classifier, Adaboost Classifier, Gaussian Naive Bayes (GaussianNB), Gradient Boosting Classifier, Grid Search CV, Extreme Gradient Boosting (XGB) Classifier. The feature selection models used in this study are feature importance and Univariate Selection. Additionally, the proposed methods are using 10-fold cross validation to acquire the finest precision rate, as well as hyperparameter tuning in each classifier to assign the best parameters. This study used the best dataset obtained from the UCI repository. The implemented method’s performance was evaluated to determine “Accuracy”, “Sensitivity” and “Specificity”. When all techniques were compared, the “Logistic Regression Classifier” provided the best accuracy of 98.25%. As a consequence, the suggested technique outperforms existing methods since it classifies the optimal features automatically and experimental findings demonstrate the model outperformed previously published “Machine Learning” methods.