Machine Learning for Air Quality Prediction

EasyChair Preprint 15683

Abstract

With air pollution being one of the major problems for urban sustainability and human health in Bishkek, accurate prediction of air quality is crucial. This study aims to predict air quality using machine learning techniques with meteorological and pollution concentration data. Using historical dataset from 2019 to 2023, we employed the CatBoostRegressor model to predict the Air Quality Index, prioritizing features such as humidity, pressure, and historical values of pollutants. Our model demonstrated exceptional performance, achieving a lower value of RMSE and R² of 0.98 on validation dataset. Our findings support the potential of machine learning in environmental monitoring and suggest improvements in air quality awareness programs.

Keyphrases: AQI, Air quality prediction, Meteorological data, data analysis, machine learning

@booklet{EasyChair:15683,
  author    = {Arafat Amrullaev and Remudin Reshid Mekuria},
  title     = {Machine Learning for Air Quality Prediction},
  howpublished = {EasyChair Preprint 15683},
  year      = {EasyChair, 2025}}