Breast Cancer Prediction Using Classical Machine Learning Models

Machine Learning is an emerging topic of study nowadays. Due to their extensive capacity for data exploration and the discovery of illuminating patterns, machine learning, and data mining techniques have become extremely popular. Data mining techniques are also used in the healthcare industry and have proven particularly effective for cancer prognosis and prediction. Breast cancer is the second most significant factor in women's cancer-related death rates. In that situation, early detection of benign and malignant stages of breast cancer can enable the patient, and the doctor administers the appropriate treatments. In this study, a breast cancer dataset of 569 instances and 32 features were chosen to predict the stages of breast cancer. Four classification algorithms are employed: Support Vector Machine, K Nearest Neighbor, Multi layer Perceptron, and Radial Basis Neural Network. The algorithm's performance regarding Accuracy, Precision, Recall, and Roc Area are compared, and it is found that SVM outperformed the others with an accuracy rate of 98.05%. The paper also demonstrated how feature selection strategies might shorten training times before proposing a reliable and universal model for predicting the breast cancer stage.