Prediction models for ozone in metropolitan area of Mexico City based on artificial intelligence techniques
by Gong Bing; Joaquín Ordieres-Meré; Claudia Barreto Cabrera
International Journal of Information and Decision Sciences (IJIDS), Vol. 7, No. 2, 2015

Abstract: Ozone is one of the worst harmful pollutants nowadays which affects the public health, so it is necessary to predict ozone level accurately in order to prevent the public from exposing to the pollution when it exceeds the limits. This study aims to predict daily maximum ozone concentrations in the metropolitan area of Mexico City by using four individual artificial intelligence techniques: multiple linear regression, neural networks, support vector machine, random forest, and two ensemble techniques: linear ensemble and greedy ensemble. Results from the comparison among different artificial intelligence techniques clearly showed that ensemble models, especially linear ensemble model, outperformed the individual artificial intelligence techniques. Moreover, it is concluded that the performance of models is influenced by the time ahead factor for the predictors. The errors of prediction models related to the data of current day are only around 50% of ones corresponding to the data of the previous day. In addition, in order to select the input variables properly, analysis of variance (ANOVA) based on multiple linear regression models was performed. Best model prediction capability also depends on the ranges of input variables.

Online publication date: Sun, 12-Apr-2015

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Information and Decision Sciences (IJIDS):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com