Lung cancer stands as a prevalent respiratory ailment worldwide, with its incidence intricately linked to air pollution. Investigating this relationship is pivotal for implementing effective preventive strategies. Traditionally, research has relied on a simplistic causal model, assuming uniform relationships between air pollutants and lung cancer occurrence. However, this assumption falls short of reality. This study aims to enhance the predictive accuracy of d-vine copula modeling which elucidate the complex relationship between various air pollutants—such as PM \(_{2.5}\) , PM \(_{1.0}\) , CO, SO \(_2\) , and SO \(_4\) —and the occurrence of lung cancer in both genders comprehensively, by employing a decision tree-based algorithm, specifically using bagging machine learning algorithm. This method allows for the simultaneous exploration of relationships among the important variables with diverse natures and structures. The results reveal distinctive characteristics and structures in the association between air pollutants and lung cancer incidence in men and women, highlighting variations in tail dependencies. Specifically, PM \(_{1.0}\) and SO \(_4\) emerge as pivotal factors for men, whereas PM \(_{2.5}\) exerts the greatest influence on women. In our comparison, we found that employing a bagging algorithm into the d-vine copula modeling, resulted in superior accuracy compared to d-vine copula modeling without bagging. The mean absolute percentage error (MAPE) of the d-vine copula model was significantly reduced from 4.954% for men and 7.412% for women to 3.861% for men and 6.613% for women. This comprehensive approach sheds light on the complex interplay between air pollutants and lung cancer, offering valuable insights for targeted intervention and mitigation strategies.
