A Novel Predictive Model for Environmental Performance Assessment of Airport Operations Based on Decision Trees

The influence of airport operations is neglected during the environmental performance assessment (EPA) of aviation sector. Limited EPA tools such as airport sustainability reports and green building rating tools (GBRT) exist to evaluate airport operations. However, airports utilize the gaps in these tools for leveraging economic benefits without considerable environmental contribution. This research is aimed to address this gap by developing a supervised model for EPA of airport operations. A database is created with 27 features related to airport sustainability, by following a web-mining and content analysis approach in their sustainability, environment, corporate-social responsibility, and annual reports. The Classification and Regression tree (CART) model is preferred based on its higher learning accuracy with scarce data and easier interpretability of results by airport stakeholders. The CART model inferred 155 green rules with a predictive accuracy of 88.18 %. Scope 2 emissions is observed as the significant environmental feature by performing an occurrence and impact analysis. The model learnt, is tested for its dynamic prediction ability by observing new and justifiable inferences based on the impacts of COVID-19 and global environmental policies. The CART model is superior to GBRTs and hence can serve as an ideal tool for airport EPA.The influence of airport operations is neglected during the environmental performance assessment (EPA) of aviation sector. Limited EPA tools such as airport sustainability reports and green building rating tools (GBRT) exist to evaluate airport operations. However, airports utilize the gaps in these tools for leveraging economic benefits without considerable environmental contribution. This research is aimed to address this gap by developing a supervised model for EPA of airport operations. A database is created with 27 features related to airport sustainability, by following a web-mining and content analysis approach in their sustainability, environment, corporate-social responsibility, and annual reports. The Classification and Regression tree (CART) model is preferred based on its higher learning accuracy with scarce data and easier interpretability of results by airport stakeholders. The CART model inferred 155 green rules with a predictive accuracy of 88.18 %. Scope 2 emissions is observed as the significant environmental feature by performing an occurrence and impact analysis. The model learnt, is tested for its dynamic prediction ability by observing new and justifiable inferences based on the impacts of COVID-19 and global environmental policies. The CART model is superior to GBRTs and hence can serve as an ideal tool for airport EPA.