PREDICTION OF CANCER DIAGNOSIS USING LOGISTIC REGRESSION AND DECISION TREE CLASSIFIERS

Abstract

Cancer remains a leading cause of morbidity and mortality worldwide, necessitating improved diagnostic methods for early detection and prevention. This study investigates the application of machine learning techniques, specifically Logistic Regression (LR) and Decision Tree (DT) classifiers, for predicting cancer diagnosis using the Wisconsin dataset. The dataset, comprising 1,500 patient records with demographic, lifestyle, and clinical attributes, was selected due to its rich feature set and reliability. Logistic regression analysis revealed significant predictors of cancer, including age, gender, body mass index (BMI), smoking status, genetic risk, alcohol intake, and prior cancer history, while physical activity showed a protective effect. Model performance was evaluated using confusion matrices, accuracy, precision, recall, F1-score, and ROC-AUC metrics. Results indicated that LR achieved strong predictive accuracy (optimal threshold 0.4–0.5, AUC = 0.91), balancing sensitivity and specificity effectively. The DT model provided interpretable decision rules, identifying cancer history, genetic risk, BMI, and age as key discriminators. Both approaches highlight the potential of regression-based and tree-based models in supporting early cancer diagnosis. These findings demonstrate the utility of predictive modeling in clinical decision-making and underscore the importance of integrating lifestyle and genetic risk factors into diagnostic tools.