A novel neural network ensemble architecture for time series forecasting

Iffat A. Gheyas, Leslie S. Smith

Research output: Contribution to journalArticle

41 Citations (Scopus)

Abstract

We propose a novel homogeneous neural network ensemble approach called Generalized Regression Neural Network (GEFTS-GRNN) Ensemble for Forecasting Time Series, which is a concatenation of existing machine learning algorithms. GEFTS uses a dynamic nonlinear weighting system wherein the outputs from several base-level GRNNs are combined using a combiner GRNN to produce the final output. We compare GEFTS with the 11 most used algorithms on 30 real datasets. The proposed algorithm appears to be more powerful than existing ones. Unlike conventional algorithms, GEFTS is effective in forecasting time series with seasonal patterns.
Original languageEnglish
Pages (from-to)3855-3864
Number of pages10
JournalNeurocomputing
Volume74
Issue number18
DOIs
Publication statusPublished - Nov 2011
Externally publishedYes

Fingerprint

Network architecture
Time series
Neural networks
Learning algorithms
Nonlinear Dynamics
Learning systems
Nonlinear systems

Cite this

Gheyas, Iffat A. ; Smith, Leslie S. / A novel neural network ensemble architecture for time series forecasting. In: Neurocomputing. 2011 ; Vol. 74, No. 18. pp. 3855-3864.
@article{f587c9b7400c4c43b8c7c4fbf5c667cf,
title = "A novel neural network ensemble architecture for time series forecasting",
abstract = "We propose a novel homogeneous neural network ensemble approach called Generalized Regression Neural Network (GEFTS-GRNN) Ensemble for Forecasting Time Series, which is a concatenation of existing machine learning algorithms. GEFTS uses a dynamic nonlinear weighting system wherein the outputs from several base-level GRNNs are combined using a combiner GRNN to produce the final output. We compare GEFTS with the 11 most used algorithms on 30 real datasets. The proposed algorithm appears to be more powerful than existing ones. Unlike conventional algorithms, GEFTS is effective in forecasting time series with seasonal patterns.",
keywords = "Curse of dimensionality, Deseasonalization, Dynamic nonlinear weighted voting, Generalized regression neural networks, Neural network ensemble, Time series forecasting",
author = "Gheyas, {Iffat A.} and Smith, {Leslie S.}",
year = "2011",
month = "11",
doi = "10.1016/j.neucom.2011.08.005",
language = "English",
volume = "74",
pages = "3855--3864",
journal = "Neurocomputing",
issn = "0925-2312",
publisher = "Elsevier",
number = "18",

}

A novel neural network ensemble architecture for time series forecasting. / Gheyas, Iffat A.; Smith, Leslie S.

In: Neurocomputing, Vol. 74, No. 18, 11.2011, p. 3855-3864.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A novel neural network ensemble architecture for time series forecasting

AU - Gheyas, Iffat A.

AU - Smith, Leslie S.

PY - 2011/11

Y1 - 2011/11

N2 - We propose a novel homogeneous neural network ensemble approach called Generalized Regression Neural Network (GEFTS-GRNN) Ensemble for Forecasting Time Series, which is a concatenation of existing machine learning algorithms. GEFTS uses a dynamic nonlinear weighting system wherein the outputs from several base-level GRNNs are combined using a combiner GRNN to produce the final output. We compare GEFTS with the 11 most used algorithms on 30 real datasets. The proposed algorithm appears to be more powerful than existing ones. Unlike conventional algorithms, GEFTS is effective in forecasting time series with seasonal patterns.

AB - We propose a novel homogeneous neural network ensemble approach called Generalized Regression Neural Network (GEFTS-GRNN) Ensemble for Forecasting Time Series, which is a concatenation of existing machine learning algorithms. GEFTS uses a dynamic nonlinear weighting system wherein the outputs from several base-level GRNNs are combined using a combiner GRNN to produce the final output. We compare GEFTS with the 11 most used algorithms on 30 real datasets. The proposed algorithm appears to be more powerful than existing ones. Unlike conventional algorithms, GEFTS is effective in forecasting time series with seasonal patterns.

KW - Curse of dimensionality

KW - Deseasonalization

KW - Dynamic nonlinear weighted voting

KW - Generalized regression neural networks

KW - Neural network ensemble

KW - Time series forecasting

UR - http://www.scopus.com/inward/record.url?scp=80053306318&partnerID=8YFLogxK

U2 - 10.1016/j.neucom.2011.08.005

DO - 10.1016/j.neucom.2011.08.005

M3 - Article

VL - 74

SP - 3855

EP - 3864

JO - Neurocomputing

JF - Neurocomputing

SN - 0925-2312

IS - 18

ER -