Efficient Features Selection Based Breast Tumors Classification with Machine Learning

A tumor is the most dangerous word in health sciences worldwide. Hundreds of people are dying due to tumors every day. Breast tumor is one of the highest death factors among all types of tumors, especially in women. According to the literature, 40k people have died from breast tumors in a single year. This death percentage is relatively high in the last or fourth stage as compared to the early stages. That is why early stages detection using different machine learning techniques is required. The main objective is diagnosing and predicting breast tumors at early stages to overcome death rates. This research focuses on the early detection of breast tumors with the help of Machine Learning (ML) models and features selection. Moreover, this research compares the performances of the selected ML model, i.e., Gaussian Naïve Bayes, Linear Regression Classifier, AdaBoost Ensemble, and Neural Network. The experiments are done on a benchmark dataset, Wisconsin breast cancer (diagnostic). Our results show that one can easily classify and detect whether the tumor is in an early stage or not with 97.66%±1.12 accuracy.