A Hybrid Data Mining Model for Early Detection of Lung Cancer Utilizing Supervised Feature Extraction

The escalating global burden of cancer, particularly lung cancer, underscores the urgent need for innovative diagnostic tools. This chapter presents a hybrid data mining model that leverages supervised feature extraction to enhance the early detection of lung cancer. By employing algorithms such as Best First (BF) and Greedy Stepwise (GS) for feature selection, and machine learning classifiers like J48, Random Tree (RT), Logistic Regression (LR), and Bayes Net (BN), the model aims to improve diagnostic accuracy and efficiency. The experimental results highlight the superior performance of the Random Tree algorithm, achieving an impressive accuracy rate of 98.4%. The chapter also identifies alcohol consumption as the most critical feature influencing lung cancer prediction. Through a detailed analysis of various performance metrics, including accuracy, Kappa Statistic, Mean Absolute Error, and ROC-AUC, the study provides a robust framework for developing reliable predictive models. The discussion section offers insights into the strengths and limitations of different algorithms, paving the way for future research in AI-driven cancer detection.