Time series forecasting with neural networks: a comparative study using the air line data

By Julian J. Faraway, C. Chatfield. 2008

319 citations recorded by the source corpus.

This case‐study fits a variety of neural network (NN) models to the well‐known air line data and compares the resulting forecasts with those obtained from the Box–Jenkins and Holt–Winters methods. Many potential problems in fitting NN models were revealed such as the possibility that the fitting routine may not converge or may converge to a local minimum. Moreover it was found that an NN model which fits well may give poor out‐of‐sample forecasts. Thus we think it is unwise to apply NN models blindly in ‘black box’ mode as has sometimes been suggested. Rather, the wise analyst needs to use traditional modelling skills to select a good NN model, e.g. to select appropriate lagged variables as the ‘inputs’. The Bayesian information criterion is preferred to Akaike's information criterion for comparing different models. Methods of examining the response surface implied by an NN model are examined and compared with the results of alternative nonparametric procedures using generalized additive models and projection pursuit regression. The latter imposes less structure on the model and is arguably easier to understand.

Subfields: Artificial Intelligence, Management Science and Operations Research, Statistics and Probability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Management Science and Operations Research

In Statistics and Probability

Citations in the corpus

Classification-based financial markets prediction using deep neural networks (2016, neutral): “While the application of artiﬁcial neural networks (ANNs) to time series methods are well documented (Faraway and Chatﬁeld, 1998; Refenes, 1994; Trippi and DeSieno, 1992; Kaas-tra and Boyd, 1995) their proneness to over-ﬁtting, convergence problems, and diﬃculty of implementation raised concerns.”
Meta-learning for time series forecasting and forecast combination (2010, methodology): “The machine learning community mainly looks at neural networks, either using Multi-Layer-Perceptrons with time-lagged time series observations as inputs as, for example, in [49] and [16], or recurrent networks with a memory, see, for example, [26].”
Sequence classification of the limit order book using recurrent neural networks (2017, methodology): “The application of artiﬁcial neural networks (ANNs) to uni-variate ﬁnancial time series methods are well documented (Faraway and Chatﬁeld, 1998; Refenes, 1994; Trippi and DeSieno, 1992; Kaastra and Boyd, 1995).”
Avoiding Pitfalls in Neural Network Research (2007, methodology): “In-sample model selection approaches based on traditional information-based criteria such as Akaike’s information criterion (AIC) and Bayesian (BIC) or Schwarz information criterion (SIC) have been proposed and used in time series studies [21], [43].”
The use of artificial neural networks to analyze and predict alongshore sediment transport (2010, background): “There is no clear solution to this problem, which is why most authors prefer to train ANNs using different random seeds to generate initial weights and then analyze the best ANN (Faraway and Chatfield, 1998).”
Improved supply chain management based on hybrid demand forecasts (2007, methodology): “There are many publications comparing ARIMA and neural networks, both theoretically [6,20] and empirically [ 7 ,12].”
An Introductory Study on Time Series Modeling and Forecasting (2013, methodology): “ny time series such as those related to socio-economic [23] and - 22 - business show non-stationary behavior. Time series, which contain trend and seasonal patterns, are also non-stationary in nature [3, 11]. Thus from application view point ARMA models are inadequate to properly describe non-statio”