Breast Cancer Risk Detection using XGB Classification Machine Learning Technique

Breast cancer is cancer that forms in the breast tissue. Among all cancer deaths worldwide, many women are dying, majorly due to breast cancer. Approximately 10% of women in the world are getting affected by breast cancer at some point in their lives. Therefore, it has become a significant health problem that needs to be dealt with urgently. However, diagnosing breast cancer at early stages can increase the survival rates of patients as they can receive prompt treatment at the correct time. Additionally, it is essential to identify which type of tumor a patient has to avoid unnecessary treatment procedures. There are two types of cancers. One is benign, and the other is malignant. Both have different treatment methods, so identifying the type plays a significant role. However, the process is tiresome and may lead to disparities among pathologists. Machine learning procedures have been found to show high prediction accuracy. In order to decrease the mortality rate due to breast cancer, it is essential to predict at early stages. Wisconsin Breast cancer datasets are used for our study. Two classification models, i.e., XGB (Extreme Gradient Boosting) classifier algorithm and Gradient boosting Classifiers, are used to work on BC datasets. Datasets have been divided into training data set (80%) and test data set (20%). Initially, threefold validation methods are employed. A later prediction experiment was carried out using an XGB classifier and various performance evaluation metrics such as Confusion matrix, recall, precision, etc. By using the results obtained from performance metrics of algorithms, the best-suited model has been chosen. In our study, the XGB classifier showed the best prediction accuracy than the Light Gradient Boosting classifier (LGBM).