REAL TIME AIR QUALITY PREDICTION AND HEALTH ADVISORY SYSTEM USING MACHINE LEARNING

      The Real-Time Air Quality Prediction and Health Advisory System performed very well with regard to predicting values for the Air Quality Index (AQI). The results obtained from training the system with environmental data were highly accurate and consistent regardless of whether historic or real-time data was used. The system efficiently learnt the correlation of AQI levels with multiple pollutants.       Out of the various input parameters used in the analysis, it was found that PM2.5 and PM10 played the significant role in the AQI prediction process, and this can be validated from practical experience. The model was able to identify short-term as well as long-term variations in air quality data. The training as well as the validation process indicated that the convergence was stable along with low errors, and hence there is no sign of overfitting. By visualizing the predicted values of AQI, it was seen that the model follows the real-world trends, and hence is reliable. The system also assigns AQI levels of Good, Moderate, Poor, and Hazardous along with health advisories.   Despite its good performance, there might be minor discrepancies in its accuracy when subjected to abrupt environmental changes and extremely polluted environments. These shortcomings could be minimized through the addition of other data sources like weather data and IoT devices. In summary, the framework is dependable, scalable, and efficient for air quality prediction purposes

[1] Breiman, L. (2001). Random forests. Machine Learning, 45(1), 5-32. Retrieved from https://doi.org/10.1023/A:1010933404324 [2] Castelli, M., & Vanneschi, L., & Popovič, A. (2014). Predicting air quality index using support vector machines and random forests. Environmental Science and Pollution Research, 21. Retrieved from https://doi.org/10.1007/s11356-014-2782-z [3] Chen, T., & Guestrin, C. (2016). XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Retrieved from https://doi.org/10.1145/2939672.2939785 [4] Diaz-Robles, L. A., & Fu, J. S., & Reed, G. D. (2008). A hybrid ARIMA and artificial neural networks model for particulate matter forecasting in San Jose, Costa Rica. Atmospheric Environment, 42(35). Retrieved from https://doi.org/10.1016/j.atmosenv.2008.07.035 [5] Ghojogh, B., & Crowley, M. (2019). The theory behind overfitting and regularization in machine learning. arXiv preprint. Retrieved from https://arxiv.org/abs/1901.01015 [6] Gupta, S., & Christopher, S. A. (2009). Particulate matter air quality assessment using integrated surface, satellite, and meteorological products. Journal of Geophysical Research: Atmospheres, 114. Retrieved from https://doi.org/10.1029/2008JD011496 [7] Hastie, T., & Tibshirani, R., & Friedman, J. (2009). The Elements of Statistical Learning: Data Mining, Inference, and Prediction. Springer Science & Business Media. [8] Ho, T. K. (1995). Random decision forests. Proceedings of 3rd International Conference on Document Analysis and Recognition. Retrieved from https://doi.org/10.1109/ICDAR.1995.598994 [9] Kadiyala, A., & Kumar, A. (2013). Applications of soft computing techniques in air quality forecasting: A review. International Journal of Environmental Waste Management, 12(1). Retrieved from https://doi.org/10.1504/IJEWM.2013.054366 [10] Kumar, K., & Goyal, P. (2011). Forecasting of daily air quality index using artificial neural networks: A case study of Delhi. Science of the Total Environment, 409(23). Retrieved from https://doi.org/10.1016/j.scitotenv.2011.08.019