Quarterly Time-Series Forecasting With Neural Networks

Abstract

Forecasting of time series that have seasonal and other variations remains an important problem for forecasters. This paper presents a neural network (NN) approach to forecasting quarterly time series. With a large data set of 756 quarterly time series from the M3 forecasting competition, we conduct a comprehensive investigation of the effectiveness of several data preprocessing and modeling approaches. We consider two data preprocessing methods and 48 NN models with different possible combinations of lagged observations, seasonal dummy variables, trigonometric variables, and time index as inputs to the NN. Both parametric and nonparametric statistical analyses are performed to identify the best models under different circumstances and categorize similar models. Results indicate that simpler models, in general, outperform more complex models. In addition, data preprocessing especially with deseasonalization and detrending is very helpful in improving NN performance. Practical guidelines are also provided.

Data Pre-Processing for Neural Network-Based Forecasting: Does It Really Matter?

“…Our finding on the suitability of working with seasonally adjusted levels for neural network forecasting confirms previous research by Zhang and Kline (2007), Zhang and Qi (2005), Virili and Freisleben (2000) and Nelson et al (1999). The fact that we do not find evidence in favour of detrending can be explained by the fact that the data used for the analysis does not present a strong trend component.…”

“…Nevertheless, studies reach different conclusions on how to deal with seasonal time series (Hamzaçebi 2008). While Nelson et al (1999) and Zhang and Kline (2007) concluded that in order to obtain a better ANN forecasting, the seasonal effect should be removed form the raw data, Franses and Draima (1997) and Alon et al (2001) found that ANNs are capable of modelling the seasonal and trend effects in data structure without removing the seasonal effects.…”

“…Nevertheless, studies reach different conclusions on how to deal with seasonal time series (Hamzaçebi 2008). While Nelson et al (1999) and Zhang and Kline (2007) concluded that in order to obtain a better ANN forecasting, the seasonal effect should be removed form the raw data, Franses and Draima (1997) and Alon et al (2001) found that ANNs are capable of modelling the seasonal and trend effects in data structure without removing the seasonal effects.The objective of this study is to investigate the effects of data pre-processing in the forecast performance of ANNs when using seasonal time series, extending to tourist demand forecasting the results of previous research on economics. Given that univariate specifications are limited and unable to capture dynamic interrelationships between variables, we analyze whether a multivariate setting provides useful for forecasting purposes.…”

Time Series Seasonal Analysis Based on Fuzzy Transforms

“…The union of k − 1 subsets is used as training set and the other subset is used as validation set. Then we consider a fuzzy partition with n = 3 and apply Equation (17) to the training set calculating the MADMEAN index; then we apply Equation (18) to the validation test to calculate the RMSE index. We repeat this process for all the k folds, obtaining the mean MADMEAN and RMSE indexes as…”

“…The main advantage of an SVM method is that that the solution is unique and there are no risk to move towards local minima, but some problems remain as the choice of the kernel parameters which influences the structure of the feature space, affecting the final solution. Another method is based on an Artificial Neural Network (ANN) [8,[15][16][17]. The most widely used ANN architectures for forecasting problems are given by multi-layer Feed Forward Network (FNN) architectures [18,19], where the input nodes are given by the successive observations of the time series; that is, target y t is a function of the values y t−1 , y t−2 , .…”