AQI Prediction Using Human-AI Teaming

The paper aims to enhance the prediction of the Air Quality Index (AQI) by incorporating human-AI interaction in conjunction with Random Forest Regression (RFR). Although ML techniques have been applied in PA, these methods are highly susceptible to the complexity of variables in domains like air quality. To tackle this problem, the approach proposed in this paper relies on the Human-In-the-Loop approach where the confidence levels of the ML predictions are evaluated in conjunction with their accuracy. In situations where the output of the ML model is below a certain dynamic value, which is an indication that the model is not certain about the next course of action, the system reroutes the control back to a Human Expert for manual intervention. This approach is beneficial as it is more accurate than the base model when no human input is required in the algorithm’s prediction process. In addition, it also opens channels for feedback where the base model can improve itself over time through learning and modification. In the present work, we are concerned with Using a Historical Air Pollution Dataset for Oragadam, Tamil Nadu and we are particularly focusing on the year 2023. This choice is made based on the Random Forest regression’s ability to provide more stability and better capture of the non-linear nature of the data inherent in air quality. Our approach aims to address the weakness of purely automated prediction systems while addressing the strong points of both human knowledge and computational modeling to improve the accuracy of the AQI forecast. This study provides value to the enhancement of the field of collaboration between Human and Artificial Intelligence in environmental prediction with the prospect of enhancing the health of the public and the environment.