Rights statement: The final, definitive version of this article has been published in the Journal, Neurocomputing 73 (10-12), 2010, © ELSEVIER.
Submitted manuscript, 476 KB, PDF document
Accepted author manuscript
Research output: Contribution to Journal/Magazine › Journal article › peer-review
Research output: Contribution to Journal/Magazine › Journal article › peer-review
}
TY - JOUR
T1 - Feature selection for time series prediction - A combined filter and wrapper approach for neural networks
AU - Crone, Sven F.
AU - Kourentzes, Nikolaos
N1 - The final, definitive version of this article has been published in the Journal, Neurocomputing 73 (10-12), 2010, © ELSEVIER.
PY - 2010/6
Y1 - 2010/6
N2 - 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.
AB - 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.
KW - Time series prediction
KW - Forecasting
KW - Artificial neural networks
KW - Automatic model specification
KW - Feature selection
KW - Input variable selection
U2 - 10.1016/j.neucom.2010.01.017
DO - 10.1016/j.neucom.2010.01.017
M3 - Journal article
VL - 73
SP - 1923
EP - 1936
JO - Neurocomputing
JF - Neurocomputing
SN - 0925-2312
IS - 10-12
ER -