The behaviour of the multi-layer perceptron and the support vector regression learning methods in the prediction of NO and NO 2 concentrations in Szeged, Hungary

István Juhos, L. Makra, Balázs Tóth

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

The main aim of this paper is to predict NO and NO 2 concentrations 4 days in advance by comparing two artificial intelligence learning methods, namely, multi-layer perceptron and support vector machines, on two kinds of spatial embedding of the temporal time series. Hourly values of NO and NO 2 concentrations, as well as meteorological variables were recorded in a cross-road monitoring station with heavy traffic in Szeged, in order to build a model for predicting NO and NO 2 concentrations several hours in advance. The prediction of NO and NO 2 concentrations was performed partly on the basis of their past values, and partly on the basis of temperature, humidity and wind speed data. Since NO can be predicted more accurately, its values were considered primarily when forecasting NO 2. Time series prediction can be interpreted in a way that is suitable for artificial intelligence learning. Two effective learning methods, namely, multi-layer perceptron and support vector regression are used to provide efficient non-linear models for NO and NO 2 time series predictions. Multi-layer perceptron is widely used to predict these time series, but support vector regression has not yet been applied for predicting NO and NO 2 concentrations. Three commonly used linear algorithms were considered as references: 1-day persistence, average of several day persistence and linear regression. Based on the good results of the average of several day persistence, a prediction scheme was introduced, which forms weighted averages instead of simple ones. The optimization of these weights was performed with linear regression in linear case and with the learning methods mentioned in non-linear case. Concerning the NO predictions, the non-linear learning methods give significantly better predictions than the reference linear methods. In the case of NO 2, the improvement of the prediction is considerable, however, it is less notable than for NO.

Original languageEnglish
Pages (from-to)193-205
Number of pages13
JournalNeural Computing and Applications
Volume18
Issue number2
DOIs
Publication statusPublished - Feb 2009

Fingerprint

Multilayer neural networks
Time series
Linear regression
Artificial intelligence
Support vector machines
Atmospheric humidity
Monitoring

Keywords

  • Artificial neural networks
  • Forecast
  • Multi-layer perceptron
  • Support vector machines
  • Support vector regression

ASJC Scopus subject areas

  • Artificial Intelligence
  • Software

Cite this

@article{df03b0df783c4f50b08c4947952df729,
title = "The behaviour of the multi-layer perceptron and the support vector regression learning methods in the prediction of NO and NO 2 concentrations in Szeged, Hungary",
abstract = "The main aim of this paper is to predict NO and NO 2 concentrations 4 days in advance by comparing two artificial intelligence learning methods, namely, multi-layer perceptron and support vector machines, on two kinds of spatial embedding of the temporal time series. Hourly values of NO and NO 2 concentrations, as well as meteorological variables were recorded in a cross-road monitoring station with heavy traffic in Szeged, in order to build a model for predicting NO and NO 2 concentrations several hours in advance. The prediction of NO and NO 2 concentrations was performed partly on the basis of their past values, and partly on the basis of temperature, humidity and wind speed data. Since NO can be predicted more accurately, its values were considered primarily when forecasting NO 2. Time series prediction can be interpreted in a way that is suitable for artificial intelligence learning. Two effective learning methods, namely, multi-layer perceptron and support vector regression are used to provide efficient non-linear models for NO and NO 2 time series predictions. Multi-layer perceptron is widely used to predict these time series, but support vector regression has not yet been applied for predicting NO and NO 2 concentrations. Three commonly used linear algorithms were considered as references: 1-day persistence, average of several day persistence and linear regression. Based on the good results of the average of several day persistence, a prediction scheme was introduced, which forms weighted averages instead of simple ones. The optimization of these weights was performed with linear regression in linear case and with the learning methods mentioned in non-linear case. Concerning the NO predictions, the non-linear learning methods give significantly better predictions than the reference linear methods. In the case of NO 2, the improvement of the prediction is considerable, however, it is less notable than for NO.",
keywords = "Artificial neural networks, Forecast, Multi-layer perceptron, Support vector machines, Support vector regression",
author = "Istv{\'a}n Juhos and L. Makra and Bal{\'a}zs T{\'o}th",
year = "2009",
month = "2",
doi = "10.1007/s00521-007-0171-1",
language = "English",
volume = "18",
pages = "193--205",
journal = "Neural Computing and Applications",
issn = "0941-0643",
publisher = "Springer London",
number = "2",

}

TY - JOUR

T1 - The behaviour of the multi-layer perceptron and the support vector regression learning methods in the prediction of NO and NO 2 concentrations in Szeged, Hungary

AU - Juhos, István

AU - Makra, L.

AU - Tóth, Balázs

PY - 2009/2

Y1 - 2009/2

N2 - The main aim of this paper is to predict NO and NO 2 concentrations 4 days in advance by comparing two artificial intelligence learning methods, namely, multi-layer perceptron and support vector machines, on two kinds of spatial embedding of the temporal time series. Hourly values of NO and NO 2 concentrations, as well as meteorological variables were recorded in a cross-road monitoring station with heavy traffic in Szeged, in order to build a model for predicting NO and NO 2 concentrations several hours in advance. The prediction of NO and NO 2 concentrations was performed partly on the basis of their past values, and partly on the basis of temperature, humidity and wind speed data. Since NO can be predicted more accurately, its values were considered primarily when forecasting NO 2. Time series prediction can be interpreted in a way that is suitable for artificial intelligence learning. Two effective learning methods, namely, multi-layer perceptron and support vector regression are used to provide efficient non-linear models for NO and NO 2 time series predictions. Multi-layer perceptron is widely used to predict these time series, but support vector regression has not yet been applied for predicting NO and NO 2 concentrations. Three commonly used linear algorithms were considered as references: 1-day persistence, average of several day persistence and linear regression. Based on the good results of the average of several day persistence, a prediction scheme was introduced, which forms weighted averages instead of simple ones. The optimization of these weights was performed with linear regression in linear case and with the learning methods mentioned in non-linear case. Concerning the NO predictions, the non-linear learning methods give significantly better predictions than the reference linear methods. In the case of NO 2, the improvement of the prediction is considerable, however, it is less notable than for NO.

AB - The main aim of this paper is to predict NO and NO 2 concentrations 4 days in advance by comparing two artificial intelligence learning methods, namely, multi-layer perceptron and support vector machines, on two kinds of spatial embedding of the temporal time series. Hourly values of NO and NO 2 concentrations, as well as meteorological variables were recorded in a cross-road monitoring station with heavy traffic in Szeged, in order to build a model for predicting NO and NO 2 concentrations several hours in advance. The prediction of NO and NO 2 concentrations was performed partly on the basis of their past values, and partly on the basis of temperature, humidity and wind speed data. Since NO can be predicted more accurately, its values were considered primarily when forecasting NO 2. Time series prediction can be interpreted in a way that is suitable for artificial intelligence learning. Two effective learning methods, namely, multi-layer perceptron and support vector regression are used to provide efficient non-linear models for NO and NO 2 time series predictions. Multi-layer perceptron is widely used to predict these time series, but support vector regression has not yet been applied for predicting NO and NO 2 concentrations. Three commonly used linear algorithms were considered as references: 1-day persistence, average of several day persistence and linear regression. Based on the good results of the average of several day persistence, a prediction scheme was introduced, which forms weighted averages instead of simple ones. The optimization of these weights was performed with linear regression in linear case and with the learning methods mentioned in non-linear case. Concerning the NO predictions, the non-linear learning methods give significantly better predictions than the reference linear methods. In the case of NO 2, the improvement of the prediction is considerable, however, it is less notable than for NO.

KW - Artificial neural networks

KW - Forecast

KW - Multi-layer perceptron

KW - Support vector machines

KW - Support vector regression

UR - http://www.scopus.com/inward/record.url?scp=59449098688&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=59449098688&partnerID=8YFLogxK

U2 - 10.1007/s00521-007-0171-1

DO - 10.1007/s00521-007-0171-1

M3 - Article

VL - 18

SP - 193

EP - 205

JO - Neural Computing and Applications

JF - Neural Computing and Applications

SN - 0941-0643

IS - 2

ER -