Home > Research > Publications & Outputs > The accuracy of a procedural approach to specif...

Electronic data

View graph of relations

The accuracy of a procedural approach to specifying feedforward neural networks for forecasting

Research output: Working paper

Published

Standard

The accuracy of a procedural approach to specifying feedforward neural networks for forecasting. / Fildes, R A; Liao, K P.

Lancaster University : The Department of Management Science, 2003. (Management Science Working Paper Series).

Research output: Working paper

Harvard

Fildes, RA & Liao, KP 2003 'The accuracy of a procedural approach to specifying feedforward neural networks for forecasting' Management Science Working Paper Series, The Department of Management Science, Lancaster University.

APA

Fildes, R. A., & Liao, K. P. (2003). The accuracy of a procedural approach to specifying feedforward neural networks for forecasting. (Management Science Working Paper Series). The Department of Management Science.

Vancouver

Fildes RA, Liao KP. The accuracy of a procedural approach to specifying feedforward neural networks for forecasting. Lancaster University: The Department of Management Science. 2003. (Management Science Working Paper Series).

Author

Fildes, R A ; Liao, K P. / The accuracy of a procedural approach to specifying feedforward neural networks for forecasting. Lancaster University : The Department of Management Science, 2003. (Management Science Working Paper Series).

Bibtex

@techreport{5ae75750424147c3b329d2e48db747d2,
title = "The accuracy of a procedural approach to specifying feedforward neural networks for forecasting",
abstract = "The comparative accuracy of feedforward neural networks (NN) when applied to time series forecasting problems remains uncertain. This is because most studies suffer from either of two defects - they choose the NN from a wide range of alternatives in order to present the forecast accuracy results in the best light, or they do not compare the results with suitable benchmarks. In order to overcome both these objections this paper proposes an objective procedure for specifying a feedforward neural network models and evaluates its effectiveness by examining its forecasting performance compared with established benchmarks. After the selection of input nodes based on cross-validation, a 3-Stage procedure is proposed here which consists of sequentially selecting first the learning rate followed by the number of nodes and the initial weights. This paper shows that neural networks only perform robustly if they are built by considering these three factors jointly. In an empirical demonstration of the strength of the approach, those neural network models, built by considering all three factors, performed better than other competitive statistical methods when evaluated rigorously on a standard test data set.",
keywords = "backpropagation, neural networks, comparative forecasting accuracy, model specification",
author = "Fildes, {R A} and Liao, {K P}",
year = "2003",
language = "English",
series = "Management Science Working Paper Series",
publisher = "The Department of Management Science",
type = "WorkingPaper",
institution = "The Department of Management Science",

}

RIS

TY - UNPB

T1 - The accuracy of a procedural approach to specifying feedforward neural networks for forecasting

AU - Fildes, R A

AU - Liao, K P

PY - 2003

Y1 - 2003

N2 - The comparative accuracy of feedforward neural networks (NN) when applied to time series forecasting problems remains uncertain. This is because most studies suffer from either of two defects - they choose the NN from a wide range of alternatives in order to present the forecast accuracy results in the best light, or they do not compare the results with suitable benchmarks. In order to overcome both these objections this paper proposes an objective procedure for specifying a feedforward neural network models and evaluates its effectiveness by examining its forecasting performance compared with established benchmarks. After the selection of input nodes based on cross-validation, a 3-Stage procedure is proposed here which consists of sequentially selecting first the learning rate followed by the number of nodes and the initial weights. This paper shows that neural networks only perform robustly if they are built by considering these three factors jointly. In an empirical demonstration of the strength of the approach, those neural network models, built by considering all three factors, performed better than other competitive statistical methods when evaluated rigorously on a standard test data set.

AB - The comparative accuracy of feedforward neural networks (NN) when applied to time series forecasting problems remains uncertain. This is because most studies suffer from either of two defects - they choose the NN from a wide range of alternatives in order to present the forecast accuracy results in the best light, or they do not compare the results with suitable benchmarks. In order to overcome both these objections this paper proposes an objective procedure for specifying a feedforward neural network models and evaluates its effectiveness by examining its forecasting performance compared with established benchmarks. After the selection of input nodes based on cross-validation, a 3-Stage procedure is proposed here which consists of sequentially selecting first the learning rate followed by the number of nodes and the initial weights. This paper shows that neural networks only perform robustly if they are built by considering these three factors jointly. In an empirical demonstration of the strength of the approach, those neural network models, built by considering all three factors, performed better than other competitive statistical methods when evaluated rigorously on a standard test data set.

KW - backpropagation

KW - neural networks

KW - comparative forecasting accuracy

KW - model specification

M3 - Working paper

T3 - Management Science Working Paper Series

BT - The accuracy of a procedural approach to specifying feedforward neural networks for forecasting

PB - The Department of Management Science

CY - Lancaster University

ER -