Abstract
This study introduces a novel hybrid model to enhance the prediction of benzene concentrations in three industrial regions in Kuwait, utilizing air quality data from 2022 to 2024. The hybrid model, developed through stacking techniques, integrates multiple ML algorithms to employ their collective strengths. The initial analysis involved examining pollutant trends and correlations among benzene, toluene, ethylbenzene, and xylenes (BTEX) compounds. We applied more than ten individual machine learning models to predict benzene levels. We then applied a hyperparameter, tuning the hybrid model to further enhance its prediction performance. By combining these models, the hybrid approach demonstrated superior predictive performance, evaluated using R-squared and mean squared error metrics. The results underscore the effectiveness of the hybrid model in providing accurate benzene concentration prediction, offering valuable insights for air quality management and pollution control in industrial regions.

This publication has 17 references indexed in Scilit: