Respiratory disease classification using selected data mining techniques

Abstract

Lung cancer, known for its high mortality rate, continues to claim numerous lives worldwide. Early detection has proven to offer significant advantages, substantially improving the prospects for successful treatment, medication, and the healing process. Despite various classification methods used to identify certain illnesses, their accuracy has often been suboptimal. In this paper, we employ Linear Discriminant Analysis (LDA) as a classifier and dimensionality reduction model to enhance the predictive accuracy of lung cancer presence. This study aims to predict the occurrence of lung cancer by utilizing a set of predictor variables, including gender, age, allergy, swallowing difficulty, coughing, fatigue, alcohol consumption, wheezing, shortness of breath, yellowish finger, chronic disease, smoking, chest pain, anxiety, and peer pressure. The goal is to enable early diagnosis, leading to timely and effective interventions. The results of our investigation demonstrate that LDA achieves an impressive accuracy rate of 92.2% in predicting lung cancer presence, surpassing the performance of the C4.5 and Naïve Bayes classifiers. This finding underscores the potential of LDA as a valuable tool for the early detection of lung cancer, ultimately contributing to improved patient outcomes. Through the utilization of LDA, we hope to advance the field of medical diagnostics and enhance the prospects for successful lung cancer management and treatment.