Publication: An Advance Ozone Concentration Prediction By Implementing Machine Learning Algorithms In Selangor And Kuala Lumpur Malaysia
Date
2019-10
Authors
Ellysia Jumin
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Malaysia is a developing country especially regions in Kuala Lumpur and Selangor.
Current generation are indulging the comfort of life from the urbanization. Although it
has positive advantageous yet somehow the disadvantageous slowly worsen life on
earth through either direct or indirect impact contributing to adverse air quality. High
level of tropospheric ozone concentration exceeding prescribed level by the department
of environmental is an unfavorable air quality [17]. While ozone formation is a complex
chemical reaction affected by many precursors, however in this study few ozone
precursors has been identified and supported by past researches. Correlation between
ozone and the precursors is based on Pearson Correlation Coefficient. Meteorology data
such as wind speed and humidity are highly correlated followed by air quality data like
nitrogen oxide, carbon monoxide, nitrogen dioxide and the least correlated Sulphur
dioxide with ozone formation. Daytime dataset from 6:00 a.m. – 6:00 p.m. provide
better model compared to 24-hour dataset for the best model Boosted Decision Tree
Regression. Neural Network Regression with Gaussian Normalizer, Linear Regression
and Neural Network with Min-Max Normalizer show over fit model. Among all three
station of studies, S2 gives best data for the model development. Best model is
successfully identified thus beneficial for any early prevention pertaining community
safety in the near future. Further study on technique or by introducing complex input
parameters can be applied to improve the model selected.
Description
FYP 2 Sem 1 2019/2020
Keywords
Ozone