Using artificial intelligence methods to understand and forecast atmospheric quality parameters

Ionnis Kyriakidis, Kostas Karatzas, George Papadourakis, Jonathan Ware

Research output: Contribution to journalArticlepeer-review


Quality of life is strongly affected by the quality of the environment. Understanding and managing urban air quality is one of themain concerns of city authorities. For this purpose, it is important to extract knowledge and to be able to model the problem underinvestigation (air pollution), in order to be able to forecasts parameters of interest (pollutant concentrations). In this paper ArtificialNeural Networks and Linear Regression models are used together with a set of mathematical tools that include Principal ComponentAnalysis and Fast Fourier Transformations for the investigation and forecasting of hourly Benzene concentrations and highest dailyeight hour mean of ozone concentrations for two locations in Athens, Greece. The methodology is evaluated for its forecasting ability.Results verify the suitability of the computational approach employed and the improvement of results in comparison to previousapproaches.
Original languageEnglish
Pages (from-to)137 - 149
Number of pages12
JournalEngineering Intelligent Systems
Issue number1/2
Publication statusPublished - 1 Mar 2012


  • air quality
  • artificial neural networks
  • principal component analysis
  • fouriertransformation
  • ozone forecasting
  • benzene forecasting
  • linear regression
  • cross-validation


Dive into the research topics of 'Using artificial intelligence methods to understand and forecast atmospheric quality parameters'. Together they form a unique fingerprint.

Cite this