Makine Öğrenme Algoritmaları ile Ozon Seviyesi Tahmini

Atmosferde bulunan ozon seviyesi tüm canlıların yaşam kalitesini etkilemek ile birlikte ölüme varan sonuçlara sebebiyet vermektedir. Ozon güneş ışınlarından gelen radyason değerinden konurulmasını sağlayan bir gazdır. Bu nedenle ozon seviyesi belirli bir eşiği aştığında risk durumu çoğalmaktadır. Makine öğrenmesi algoritmaları, uygulanacak problem hakkında yeterli verinin elde edilmesi halinde bu verilerden daha önce karşılaşmadığı durumlara karşın çıkarım yapma yeteneğine sahiptir. Bu çalışmada hibrit bir makine öğrenmesi algoritması önerilerek ozon seviyesinin tahmini ile risk oluşmadan buna engel olunması hedeflenmiştir. Önerilen hibrit model iki aşamalı olarak sonuç almaktadır. İlk aşamasında genetik algoritmalar yöntemi ile kümeleme yapılarak küme sonucu XGBoost sınıflandırıcı yöntemine giriş olarak iletilmektedir. Önerilen modelin uygulanabilir olduğunun gösterilmesi için sık kullanılan makine öğrenmesi yöntemlerinden destek vektör makineleri, random forest, çok katmanlı yapay sinir ağları ve XGBoost yöntemleri aynı probleme uygulanmıştır. 10-fold validation uygulanmasının ardından en başarılı doğruluk oranına %94 ile önerilen model ulaşmıştır.

Ozone Level Prediction with Machine Learning Algorithms

The ozone level in the atmosphere affects the quality of life of all living things as well as it can be a hazard to human health and the environment. Ozone is a gas that absorbs most of the ultraviolet radiation reaching the Earth from the Sun. However, when the ozone level exceeds a certain threshold, risks would be exacerbated. Using machine learning algorithms can help to reduce risks, making inferences from earlier obtained data even for situations, which have not encountered before. In this study, a two-phased hybrid machine learning algorithm is proposed. It helps to predict the ozone level prospectively and reduce the risks. In the first stage, clustering is made with the method of genetic algorithms and the clustering result is transmitted as an introduction to the XGBoost classifier method. To check that the proposed model is applicable, support vector machine, random forest, multi-layered neural networks and XGBoost methods, which are among the frequently used machine learning methods, have been applied and the results were compared. After the 10-fold validation applied, the proposed model reached the most successful accuracy rate with 94%.

PDF

___

[1] S. I. V. Sousa, M. C. M. Alvim-Ferraz and F. G. Martins, "Health effects of ozone focusing on childhood asthma: What is now known – a review from an epidemiological point of view" Chemosphere, vol. 90, pp. 2051-2058, June 2013.
[2] Y. C. Wang, F. C. Sung, Y. J. Chen, C. P. Cheng and Y. K. Lin, "Effects of extreme temperatures, fine particles and ozone on hourly ambulance dispatches" Science of the Total Environment, vol. 765, April 2021.
[3] Y. C. Wang, F. C. Sung, Y. J. Chen, C. P. Cheng and Y. K. Lin, "Ozone pollution in China: Background and transboundary contributions to ozone concentration & related health effects across the country" Science of the Total Environment, vol. 761, pp. 144131, March 2021.
[4] M. Pascal, V. Wagner, A. Alari, M. Corso and A. La Tertre, "Extreme heat and acute air pollution episodes: A need for joint public health warnings?" Atmospheric Environment, vol. 249, March 2021.
[5] J. Bosch, S. Elvira, C. Sausor, J. Bielby, I. F. Gonzalez and V. Bermejo, "Increased tropospheric ozone levels enhance pathogen infection levels of amphibians" Science of the Total Environment, vol. 759, March 2021.
[6] R. Long, A. Whitehill, A. Habel, S. Urbanski, H. Halliday, M. Colon, S. Kaushik and M. Landis, "Comparison of ozone measurement methods in biomass burning smoke: an evaluation under field and laboratory conditions" Atmospheric Measurement Techniques, vol. 14, pp. 1783-1800, March 2021.
[7] A. Gao, J. Wang, J. Luo, A. Li, K. Chen, P. Wang, Y. Wang, J. Li, J. Hu and H. Zhang, "Temporal variation of PM2.5-associated health effects in Shijiazhuang, Hebei" Frontiers of Environmental Science & Engineering, vol. 15, pp. 1783-1800, December 2020.
[8] A. Vellido, "The importance of interpretability and visualization in machine learning for applications in medicine and health care" Neural Computing & Applications, vol. 32, pp. 18069-18083, December 2020.
[9] H. Zhu, S. Thedoridis, K. Lam and H. Lung, “A new method for ozone depletion detection over antarctica by deep convolutional neural network” in Proc. of the 2020 IEEE 5th International Conference on Signal and Image Processing, ICSIP 2020, pp. 506-511, October 23-25, 2020.
[10] R. V. Gagliardi and C. Andenna, “A machine learning approach to investigate the surface ozone behavior” Atmosphere, vol. 11, pp. 1173, October 2020.
[11] J. Ordieres-Meré, J. Ouarzazi, B. El Johra, and B. Gong, “Predicting ground level ozone in Marrakesh by machine-learning techniques” Journal of Environmental Informatics, vol. 36, pp. 93-106, December 2020.
[12] R. Liu, Z. Ma, Y. Liu, Y. Shao, W. Zhao and J. Bi “Spatiotemporal distributions of surface ozone levels in China from 2005 to 2017: A machine learning approach” Environmental International, vol. 142, September 2020.
[13] H. W. Wang, X. B. Li, D. Wang, J. Zhao, H. D. He, and Z. R. Peng, “Regional prediction of ground- level ozone using a hybrid sequence-to-sequence deep learning approach” Journal of Cleaner Production, vol. 253, April 2020.
[14] M. K. AlOmar, M. M. Hameed and M. A. AlSaadi, “Multi hours ahead prediction of surface ozone gas concentration: Robust artificial intelligence approach” Atmospheric Pollution Research, vol. 11(9), pp. 1572-1587, September 2020.
[15] E. Jumin, N. Zaini, A. N. Ahmed, S. Abdullah, M. İsmail, M.Sherif, A. Sefelnasr and A. El-Shafie “Machine learning versus linear regression modelling approach for accurate ozone concentrations prediction” Engineering Applications of Computational Fluid Mechanics, vol. 14, May 2020.
[16] G. L. Watson, D. Telesca, C. E. Reid, G. G. Pfister and M. Jerrett, “Machine learning models accurately predict ozone exposure during wildfire events” Environmental Pollution, vol. 254, November 2019.
[17] O. A. Ghoneim, H. Doreswamy and B. Manjunatha, "Forecasting of ozone concentration in smart city using deep learning," in 2017 International Conference on Advances in Computing, Communications and Informatics (ICACCI), ICACCI 2017, Udupi, India, September 13-16, 2017.
[18] G. L. Watson, D. Telesca, C. E. Reid, G. G. Pfister and M. Jerrett, “Forecasting skewed biased stochastic ozone days : Analyses, solutions and beyond” Knowledge and Information Systems, vol. 14, pp. 299-326, July 2007.
[19] UCI Machine Learning Repository, Ozone Level Detection Dataset, https://archive.ics.uci.edu/ml/data setsozone+level+detection [Accessed: March. 2021].
[20] M. J. Wang and H. L. Chen, “Chaotic multi- swarm whale optimizer boosted support vector machine for medical diagnosis” Applied Soft Computing, vol. 88, March 2020.
[21] W. Chen, Y. Li, W. F. Xue, H. Shahabi, S. J. Li, H. Y. Hong, X. J. Wang, H. Y. Bian, S. Zhang, B. Pradhan and B. B. Ahmad, “Modeling flood susceptibility using data-driven approaches of naive Bayes tree, alternating decision tree, and random forest methods” Science of the Total Environment, vol. 701, January 2020.
[22] J. P. Lin, C. W. Qi, H. L. Wan, J. Y. Min, J. J. Chen, K. Zhang and L. Zhang, “Prediction of cross- tension strength of self-piercing riveted joints using finite element simulation and XGBoost algorithm” Chinese Journal of Mechanical Engineering, vol. 34, December 2021.
[23] M. J. Wang and H. L. Chen, “An improved grey wolf optimization-based learning of artificial neural network for medical data classification” Journal of Information and Communication Technology, vol. 20, pp. 213-248, April 2021.
[24] K. Jebari and M. Madiafi, “Selection methods for genetic algorithms” International Journal of Emerging Sciences, vol. 3, pp. 333-344, December 2013.
[25] M. Üstüner, S. Abdikan, G. Bilgin and F. B.Şanlı, “Crop Classification Using Light Gradient Boosting Machines” Turkish Journal of Remote Sensing and GIS, vol. 1, pp. 97-105, September 2020.