Detection of Breast Cancer Using Machine Learning Algorithms: A Study on Logistic Regression, K-Nearest Neighbors, and Decision Trees

Abstract

Background

Breast cancer is a malignant neoplasm that originates from breast cells. It is a major issue that affects both developed and developing nations. Prompt identification considerably increases the chance of effective therapy. Regular tests and public awareness campaigns are very helpful in the early identification of breast cancer. Machine learning offers potential in breast cancer prediction via the identification of hidden patterns in data. Several ML prediction models have been studied to accurately predict breast cancer.

Methods

In this study, three classifiers LR, KNN and DT were examined to detect malignant and benign samples in the Wisconsin Breast Cancer Dataset (WBCD). Model performance is evaluated using evaluation measures including recall, F1-score, accuracy, precision, and error rate.

Results

Early breast cancer detection is facilitated by the combination of many risk factors. It facilitates the development of necessary care plans. Managing, storing, and gathering a variety of data is essential. Multi-factor intelligent systems improve breast cancer prediction. They work well in the treatment of disease. The Logistic Regression model had the highest accuracy of 99.12% and demonstrated strong predictive capabilities. K-Nearest Neighbors (KNN) had slightly lower accuracy at 96.49%. It still shows promising results in classifying breast cancer instances. Decision Tree, with an accuracy of 92.98%, showed competitive precision, recall, and F1-score metrics for both classes. It indicates strong performance in classifying breast cancer instances based on provided features.