|
Activity Number:
|
352
- SPEED: Biometrics and Environmental Statistics Part 2
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 9, 2022 : 2:00 PM to 2:45 PM
|
|
Sponsor:
|
Section on Statistics and the Environment
|
|
Abstract #323819
|
|
|
Title:
|
The Performance of Boosting Methods on Air Quality Data
|
|
Author(s):
|
Betül Kan-K?l?nç*
|
|
Companies:
|
Eski?ehir Technical University
|
|
Keywords:
|
outlier;
boosting methods;
correlation;
climate
|
|
Abstract:
|
The concentrations of air pollution have become more important due to their harmful effects on human health. Polices and air quality regimes are aimed at providing the lowest target concentrations of possible pollutants. The aim of this study is to investigate the significant variables such as climate effects and human factors by using daily patterns of PM10, NOX and NO2 and to provide meaningful predictive models for air quality control in Turkey. Novel boosting modeling methods are applied to a real data set when atypical observations and high correlations exist in the data set. The models were fitted by Gradient Boosting methods; XGBoost, catBoost and lightgbm and the results are compared. The prediction error is measured by root mean square error and mean absolute error. Results might be useful for researchers studying air pollution and can be used for the development of health policy. All analyses were conducted using RStudio Software.
|
Authors who are presenting talks have a * after their name.
