Automatic ARMA identification using neural networks and the extended sample autocorrelation function: a reevaluation

Tim Chenoweth, Robert Hubata, Robert St Louis

Research output: Contribution to journalArticle

14 Citations (Scopus)

Abstract

Recently, several researchers have attempted to use neural network approaches in conjunction with the extended sample autocorrelation function (ESACF) to automatically identify ARMA models. The work to date appears promising, but generalizations are limited by the fact that the test and training sets for the neural networks were generated from random perturbations of prototype ESACF tables. This paper develops test and training sets by varying the parameters of actual ARMA processes. The results show that the ability of neural networks to accurately identify the order of an ARMA (p,q) model from its transformed ESACF is much lower than reported by previous researchers, and is especially low for time series with fewer than 100 observations.

Original languageEnglish (US)
Pages (from-to)21-30
Number of pages10
JournalDecision Support Systems
Volume29
Issue number1
DOIs
StatePublished - Jul 2000

Fingerprint

Autocorrelation
Research Personnel
Neural networks
Time series
Autoregressive moving average
Neural Networks

ASJC Scopus subject areas

  • Management Information Systems
  • Information Systems
  • Information Systems and Management

Cite this

Automatic ARMA identification using neural networks and the extended sample autocorrelation function : a reevaluation. / Chenoweth, Tim; Hubata, Robert; St Louis, Robert.

In: Decision Support Systems, Vol. 29, No. 1, 07.2000, p. 21-30.

Research output: Contribution to journalArticle

@article{ac4977eca26b43c49c318bbfc672ed69,
title = "Automatic ARMA identification using neural networks and the extended sample autocorrelation function: a reevaluation",
abstract = "Recently, several researchers have attempted to use neural network approaches in conjunction with the extended sample autocorrelation function (ESACF) to automatically identify ARMA models. The work to date appears promising, but generalizations are limited by the fact that the test and training sets for the neural networks were generated from random perturbations of prototype ESACF tables. This paper develops test and training sets by varying the parameters of actual ARMA processes. The results show that the ability of neural networks to accurately identify the order of an ARMA (p,q) model from its transformed ESACF is much lower than reported by previous researchers, and is especially low for time series with fewer than 100 observations.",
author = "Tim Chenoweth and Robert Hubata and {St Louis}, Robert",
year = "2000",
month = "7",
doi = "10.1016/S0167-9236(00)00058-0",
language = "English (US)",
volume = "29",
pages = "21--30",
journal = "Decision Support Systems",
issn = "0167-9236",
publisher = "Elsevier",
number = "1",

}

TY - JOUR

T1 - Automatic ARMA identification using neural networks and the extended sample autocorrelation function

T2 - a reevaluation

AU - Chenoweth, Tim

AU - Hubata, Robert

AU - St Louis, Robert

PY - 2000/7

Y1 - 2000/7

N2 - Recently, several researchers have attempted to use neural network approaches in conjunction with the extended sample autocorrelation function (ESACF) to automatically identify ARMA models. The work to date appears promising, but generalizations are limited by the fact that the test and training sets for the neural networks were generated from random perturbations of prototype ESACF tables. This paper develops test and training sets by varying the parameters of actual ARMA processes. The results show that the ability of neural networks to accurately identify the order of an ARMA (p,q) model from its transformed ESACF is much lower than reported by previous researchers, and is especially low for time series with fewer than 100 observations.

AB - Recently, several researchers have attempted to use neural network approaches in conjunction with the extended sample autocorrelation function (ESACF) to automatically identify ARMA models. The work to date appears promising, but generalizations are limited by the fact that the test and training sets for the neural networks were generated from random perturbations of prototype ESACF tables. This paper develops test and training sets by varying the parameters of actual ARMA processes. The results show that the ability of neural networks to accurately identify the order of an ARMA (p,q) model from its transformed ESACF is much lower than reported by previous researchers, and is especially low for time series with fewer than 100 observations.

UR - http://www.scopus.com/inward/record.url?scp=0033738757&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0033738757&partnerID=8YFLogxK

U2 - 10.1016/S0167-9236(00)00058-0

DO - 10.1016/S0167-9236(00)00058-0

M3 - Article

AN - SCOPUS:0033738757

VL - 29

SP - 21

EP - 30

JO - Decision Support Systems

JF - Decision Support Systems

SN - 0167-9236

IS - 1

ER -