Feature selection for time series prediction - A combined filter and wrapper approach for neural networks

Authors:
Sven F. Crone;Nikolaos Kourentzes
Affiliations:
Lancaster University Management School, Department of Management Science, Centre for Forecasting, Bailrigg campus, Lancaster LA1 4YX, UK;Lancaster University Management School, Department of Management Science, Centre for Forecasting, Bailrigg campus, Lancaster LA1 4YX, UK
Venue:
Neurocomputing
Year:
2010

Citing 16
Cited 10

Multilayer feedforward networks are universal approximators

Neural Networks
Approximation capabilities of multilayer feedforward networks

Neural Networks
Neural network models for time series forecasts

Management Science
Wrappers for feature subset selection

Artificial Intelligence - Special issue on relevance
Model selection in neural networks

Neural Networks
The econometric analysis of seasonal time series

The econometric analysis of seasonal time series
Neural Networks: A Comprehensive Foundation

Neural Networks: A Comprehensive Foundation
Neural Networks for Pattern Recognition

Neural Networks for Pattern Recognition
Time Series Analysis: Forecasting and Control

Time Series Analysis: Forecasting and Control
How to Train Neural Networks

Neural Networks: Tricks of the Trade, this book is an outgrowth of a 1996 NIPS workshop
Customer Targeting: A Neural Network Approach Guided by Genetic Algorithms

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

Computers and Operations Research
Optimization-based feature selection with adaptive instance sampling

Computers and Operations Research
Methodology for long-term prediction of time series

Neurocomputing
OP-ELM: Theory, Experiments and a Toolbox

ICANN '08 Proceedings of the 18th international conference on Artificial Neural Networks, Part I
Long-Term prediction of time series using state-space models

ICANN'06 Proceedings of the 16th international conference on Artificial Neural Networks - Volume Part II

Editorial: European Symposium on Times Series Prediction

Neurocomputing
Vehicle kinematics ahead prediction for localized impact by the means of autoregressive model with exogenous input

ECC'11 Proceedings of the 5th European conference on European computing conference
Feature selection for unlabeled data

ICSI'11 Proceedings of the Second international conference on Advances in swarm intelligence - Volume Part II
Performance evaluation of feedforward neural networks for modeling a vehicle to pole central collision

GEMESED'11 Proceedings of the 4th WSEAS international conference on Energy and development - environment - biomedicine
Application of type-2 neuro-fuzzy modeling in stock price prediction

Applied Soft Computing
Signal reconstruction, modeling and simulation of a vehicle full-scale crash test based on Morlet wavelets

Neurocomputing
Questionnaires-based skin attribute prediction using Elman neural network

Neurocomputing
Power quality disturbances classification based on S-transform and probabilistic neural network

Neurocomputing
Evolutionary support vector machines for time series forecasting

ICANN'12 Proceedings of the 22nd international conference on Artificial Neural Networks and Machine Learning - Volume Part II
Forecasting seasonal time series with computational intelligence: On recent methods and the potential of their combinations

Expert Systems with Applications: An International Journal

Quantified Score

Hi-index	0.01

Visualization

Abstract

Modelling artificial neural networks for accurate time series prediction poses multiple challenges, in particular specifying the network architecture in accordance with the underlying structure of the time series. The data generating processes may exhibit a variety of stochastic or deterministic time series patterns of single or multiple seasonality, trends and cycles, overlaid with pulses, level shifts and structural breaks, all depending on the discrete time frequency in which it is observed. For heterogeneous datasets of time series, such as the 2008 ESTSP competition, a universal methodology is required for automatic network specification across varying data patterns and time frequencies. We propose a fully data driven forecasting methodology that combines filter and wrapper approaches for feature selection, including automatic feature evaluation, construction and transformation. The methodology identifies time series patterns, creates and transforms explanatory variables and specifies multilayer perceptrons for heterogeneous sets of time series without expert intervention. Examples of the valid and reliable performance in comparison to established benchmark methods are shown for a set of synthetic time series and for the ESTSP'08 competition dataset, where the proposed methodology obtained second place.