Comparative Analysis of Unsupervised Machine Learning Models on Early Detection of Breast Cancer

Abstract

Breast cancer is currently one of the ailments that strikes women. The most frequently used in the treatment and recovery of breast cancer is mammography, which is thought to be precise and prompt in its detection rate, but this tool is totally inadequate for prediction. Machine learning techniques can be used as tools for medical professionals that will enable quick and accurate identification and diagnosis of breast cancer. The use of Machine Learning (ML) in the creation of the best model for prediction will raise the likelihood that breast cancer patients will survive. In this study, K-Nearest Neighbor (KNN), K-Means (KM), and Hierarchical Clustering (HC) were the three different Unsupervised Machine Learning (UML) techniques evaluated with accuracy, f1-score, precision, and recall, and it was observed that the K-Nearest Neighbor (KNN) model had the best accuracy of 93% and was recommended for identifying breast cancer. All of the models mentioned previously were put to the test over the demonstration in a Python environment, validated with locally sourced data from Ekiti State Teaching Hospital, Ado Ekiti, and appraised using standard metrics. When comparing this study's evaluation technique to traditional methods, analysis showed that this study had a more reliable and sustained ability to predict breast cancer in women. It is thought that by enabling medical practitioners to anticipate and make the necessary plans, this technology would aid in the effective management of patients with breast cancer.