Optimization of neuro-coefficient smooth transition autoregressive models using differential evolution

Christoph Bergmeir, Isaac Triguero, Francisco Velasco, José Manuel Benítez

Research output: Chapter in Book/Report/Conference proceedingConference PaperResearchpeer-review

Abstract

This paper presents a procedure for parameter estimation of the neuro-coefficient smooth transition autoregressive model, substituting the combination of grid search and local search of the original proposal of Medeiros and Veiga (2005, IEEE Trans. NN, 16(1):97-113) with a differential evolution scheme. The purpose of this novel fitting procedure is to obtain more accurate models under preservation of the most important model characteristics. These are, firstly, that the models are built using an iterative approach based on statistical tests, and therewith have a mathematically sound construction procedure. And secondly, that the models are interpretable in terms of fuzzy rules. The proposed procedure has been tested empirically by applying it to different real-world time series. The results indicate that, in terms of accuracy, significantly improved models can be achieved, so that accuracy of the resulting models is comparable to other standard time series forecasting methods.

Original languageEnglish
Title of host publicationHybrid Artificial Intelligent Systems - 7th International Conference, HAIS 2012, Proceedings
PublisherSpringer-Verlag London Ltd.
Pages464-473
Number of pages10
Volume7208 LNAI
EditionPART 1
ISBN (Print)9783642289415
DOIs
Publication statusPublished - 2012
Externally publishedYes
EventInternational Conference on Hybrid Artificial Intelligence Systems 2012 - Salamanca, Spain
Duration: 28 Mar 201230 Mar 2012
Conference number: 7th

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NumberPART 1
Volume7208 LNAI
ISSN (Print)03029743
ISSN (Electronic)16113349

Conference

ConferenceInternational Conference on Hybrid Artificial Intelligence Systems 2012
Abbreviated titleHAIS 2012
CountrySpain
CitySalamanca
Period28/03/1230/03/12

Keywords

  • differential evolution
  • NCSTAR
  • statistical models
  • threshold autoregressive models
  • time series

Cite this

Bergmeir, C., Triguero, I., Velasco, F., & Benítez, J. M. (2012). Optimization of neuro-coefficient smooth transition autoregressive models using differential evolution. In Hybrid Artificial Intelligent Systems - 7th International Conference, HAIS 2012, Proceedings (PART 1 ed., Vol. 7208 LNAI, pp. 464-473). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7208 LNAI, No. PART 1). Springer-Verlag London Ltd.. https://doi.org/10.1007/978-3-642-28942-2_42
Bergmeir, Christoph ; Triguero, Isaac ; Velasco, Francisco ; Benítez, José Manuel. / Optimization of neuro-coefficient smooth transition autoregressive models using differential evolution. Hybrid Artificial Intelligent Systems - 7th International Conference, HAIS 2012, Proceedings. Vol. 7208 LNAI PART 1. ed. Springer-Verlag London Ltd., 2012. pp. 464-473 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 1).
@inproceedings{82b33162a68f44d1bc4432bec16ad917,
title = "Optimization of neuro-coefficient smooth transition autoregressive models using differential evolution",
abstract = "This paper presents a procedure for parameter estimation of the neuro-coefficient smooth transition autoregressive model, substituting the combination of grid search and local search of the original proposal of Medeiros and Veiga (2005, IEEE Trans. NN, 16(1):97-113) with a differential evolution scheme. The purpose of this novel fitting procedure is to obtain more accurate models under preservation of the most important model characteristics. These are, firstly, that the models are built using an iterative approach based on statistical tests, and therewith have a mathematically sound construction procedure. And secondly, that the models are interpretable in terms of fuzzy rules. The proposed procedure has been tested empirically by applying it to different real-world time series. The results indicate that, in terms of accuracy, significantly improved models can be achieved, so that accuracy of the resulting models is comparable to other standard time series forecasting methods.",
keywords = "differential evolution, NCSTAR, statistical models, threshold autoregressive models, time series",
author = "Christoph Bergmeir and Isaac Triguero and Francisco Velasco and Ben{\'i}tez, {Jos{\'e} Manuel}",
year = "2012",
doi = "10.1007/978-3-642-28942-2_42",
language = "English",
isbn = "9783642289415",
volume = "7208 LNAI",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer-Verlag London Ltd.",
number = "PART 1",
pages = "464--473",
booktitle = "Hybrid Artificial Intelligent Systems - 7th International Conference, HAIS 2012, Proceedings",
address = "Germany",
edition = "PART 1",

}

Bergmeir, C, Triguero, I, Velasco, F & Benítez, JM 2012, Optimization of neuro-coefficient smooth transition autoregressive models using differential evolution. in Hybrid Artificial Intelligent Systems - 7th International Conference, HAIS 2012, Proceedings. PART 1 edn, vol. 7208 LNAI, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 1, vol. 7208 LNAI, Springer-Verlag London Ltd., pp. 464-473, International Conference on Hybrid Artificial Intelligence Systems 2012, Salamanca, Spain, 28/03/12. https://doi.org/10.1007/978-3-642-28942-2_42

Optimization of neuro-coefficient smooth transition autoregressive models using differential evolution. / Bergmeir, Christoph; Triguero, Isaac; Velasco, Francisco; Benítez, José Manuel.

Hybrid Artificial Intelligent Systems - 7th International Conference, HAIS 2012, Proceedings. Vol. 7208 LNAI PART 1. ed. Springer-Verlag London Ltd., 2012. p. 464-473 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7208 LNAI, No. PART 1).

Research output: Chapter in Book/Report/Conference proceedingConference PaperResearchpeer-review

TY - GEN

T1 - Optimization of neuro-coefficient smooth transition autoregressive models using differential evolution

AU - Bergmeir, Christoph

AU - Triguero, Isaac

AU - Velasco, Francisco

AU - Benítez, José Manuel

PY - 2012

Y1 - 2012

N2 - This paper presents a procedure for parameter estimation of the neuro-coefficient smooth transition autoregressive model, substituting the combination of grid search and local search of the original proposal of Medeiros and Veiga (2005, IEEE Trans. NN, 16(1):97-113) with a differential evolution scheme. The purpose of this novel fitting procedure is to obtain more accurate models under preservation of the most important model characteristics. These are, firstly, that the models are built using an iterative approach based on statistical tests, and therewith have a mathematically sound construction procedure. And secondly, that the models are interpretable in terms of fuzzy rules. The proposed procedure has been tested empirically by applying it to different real-world time series. The results indicate that, in terms of accuracy, significantly improved models can be achieved, so that accuracy of the resulting models is comparable to other standard time series forecasting methods.

AB - This paper presents a procedure for parameter estimation of the neuro-coefficient smooth transition autoregressive model, substituting the combination of grid search and local search of the original proposal of Medeiros and Veiga (2005, IEEE Trans. NN, 16(1):97-113) with a differential evolution scheme. The purpose of this novel fitting procedure is to obtain more accurate models under preservation of the most important model characteristics. These are, firstly, that the models are built using an iterative approach based on statistical tests, and therewith have a mathematically sound construction procedure. And secondly, that the models are interpretable in terms of fuzzy rules. The proposed procedure has been tested empirically by applying it to different real-world time series. The results indicate that, in terms of accuracy, significantly improved models can be achieved, so that accuracy of the resulting models is comparable to other standard time series forecasting methods.

KW - differential evolution

KW - NCSTAR

KW - statistical models

KW - threshold autoregressive models

KW - time series

UR - http://www.scopus.com/inward/record.url?scp=84858788487&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-28942-2_42

DO - 10.1007/978-3-642-28942-2_42

M3 - Conference Paper

SN - 9783642289415

VL - 7208 LNAI

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 464

EP - 473

BT - Hybrid Artificial Intelligent Systems - 7th International Conference, HAIS 2012, Proceedings

PB - Springer-Verlag London Ltd.

ER -

Bergmeir C, Triguero I, Velasco F, Benítez JM. Optimization of neuro-coefficient smooth transition autoregressive models using differential evolution. In Hybrid Artificial Intelligent Systems - 7th International Conference, HAIS 2012, Proceedings. PART 1 ed. Vol. 7208 LNAI. Springer-Verlag London Ltd. 2012. p. 464-473. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 1). https://doi.org/10.1007/978-3-642-28942-2_42