Data partition and variable selection for time series prediction using wrappers

Abstract

The purpose of this paper is a comparative study of a non-exhaustive, though representative, set of methodologies already available for the partition of the training dataset in time series prediction, and also for variable selection under the wrapper paradigm. The partition policy of the training dataset and the choice of a proper set of variables for the regression vector are known to have a significant influence in the accuracy of the predictor, no matter the choice of the prediction model. However, there has been no extensive search for a figure of merit supporting a comparative analysis. Here, two partition policies, denoted sequential and random, are compared, and among the variable selection approaches using wrappers, forward selection is contrasted with sensitivity based pruning. Five real financial time series with trends and seasonality have been considered and multilayer perceptrons are adopted as the predictor. The obtained results indicate with high confidence that the rarely adopted random partition and the computationally intensive forward selection overcomes the contestants in the whole set of experiments. I. INTR O D U C T I ONArtificial Neural Networks (short, ANNs) have been successfully applied to time series prediction ([3][19] [22][23]). The great interest in ANNs reflects their potential to alleviate problems associated with traditional linear forecasting techniques, as Box and Jenkins approaches [1], and also the possibility of interpreting the prediction task as a particular instance of learning from data, a well-known research area in machine learning [2]. Seasonality in the time series is a periodic and recursive pattern produced by some highly influential factors, as climate conditions and other periodical events with economical impact, taking financial time series into account. Apart from seasonality, it may present other nonstationary behavior, denoted trends. A wide range of prediction models are applied just after all nonstationary effects have been extracted from the time series under consideration. Zhang and Qi [26] have found that ANNs are not able to effectively capture seasonal or trend variations with the unpreprocessed raw data and either detrend or deseasonalization can dramatically reduce the forecasting error. They have also indicated that simultaneously combining detrend and deseasonalization can be considered the most effective data preprocessing approach. However, other scientific papers in the literature [5][13][16][21][23][9] have indicated the opposite, i.e. ANN models perform better when the time series contains unknown seasonal and/or trend components. Sharda and Patil [21] examined 88 seasonal time series from the M-competition [17] and they have found that ANNs can effectively model seasonality, so that extracting this behavior is not necessary. Franses and Draisma [5] have found that ANNs can also detect possibly changing seasonal patterns.The absence of consensus concerning the benefits associated with the preprocessing phase, though restricted to ANN models as...

Artificial Neural Networks to Estimate the Influence of Vehicular Emission Variables on Morbidity and Mortality in the Largest Metropolis in South America

“…The Mutual Information (MI) method is a type of filter approach to subset selection. Filters are characterized by data-based methods, as the subset is chosen according to some correlation between the input candidate and the target [45,46].…”

“…The wrapper's main advantage is that the model is taken into account to analyze the influence of each input. However, to be model-dependent, its computational effort tends to be elevated, as the model has to be adjusted to each candidate subset [46].…”

Selection of Temporal Lags for Predicting Riverflow Series from Hydroelectric Plants Using Variable Selection Methods

“…The variable selection methodologies can use information available a priori, through empirical tests of trial and error, or some information criterion. Puma-Villanueva et al [47] describe a simple example of how the general process works. Consider set V that represents the space of input variables, here limited to 3.…”

“…For example, a relevant variable does not mean that the optimal subset contains it. Likewise, the inputs that belong to the optimal subset are not necessarily appropriate [47]. Guyon and Elisseeff [17] classify the VS models into embedded, wrappers, and filters, each of them having its own advantages.…”

Lag selection for time series forecasting using Particle Swarm Optimization

“…Spectral analysis [1] and autocorrelation analysis [1] are examples of filter methods. Wrappers select the features based on the performance of computational intelligent methods [19], [11]. Filters seems to be more suitable to select features for linear models prediction.…”