Feedforward versus recurrent neural networks for forecasting monthly japanese yen exchange rates

Dematos, Giovani; Boyd, Milton S.; Kermanshahi, B.; Kohzadi, Nowrouz; Kaastra, Iebeling

doi:10.1007/bf00868008

Cited by 28 publications

(13 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Latent growth curve modelling provides another useful method for analysing intervention effects on individual cases and aggregated time series . Use of artificial neural networks has also been suggested as a method for TSA which can capture non‐linearity and chaotic behaviour . Finally, periodograms and spectral analysis can be used to identify complex cycles and seasonality in the data .…”

Section: Discussionmentioning

confidence: 99%

Understanding and using time series analyses in addiction research

et al. 2019

View full text Add to dashboard Cite

Time series analyses are statistical methods used to assess trends in repeated measurements taken at regular intervals and their associations with other trends or events, taking account of the temporal structure of such data. Addiction research often involves assessing associations between trends in target variables (e.g. population cigarette smoking prevalence) and predictor variables (e.g. average price of a cigarette), known as a multiple time series design, or interventions or events (e.g. introduction of an indoor smoking ban), known as an interrupted time series design. There are many analytical tools available, each with its own strengths and limitations. This paper provides addiction researchers with an overview of many of the methods available (GLM, GLMM, GLS, GAMM, ARIMA, ARIMAX, VAR, SVAR, VECM) and guidance on when and how they should be used, sample size det ermination, reporting and interpretation. The aim is to provide increased clarity for researchers proposing to undertake these analyses concerning what is likely to be acceptable for publication in journals such as Addiction. Given the large number of choices that need to be made when setting up time series models, the guidance emphasizes the importance of pre‐registering hypotheses and analysis plans before the analyses are undertaken.

show abstract

Section: Discussionmentioning

confidence: 99%

Understanding and using time series analyses in addiction research

et al. 2019

View full text Add to dashboard Cite

show abstract

“…Therefore, in n to feedforward neural networks, the recurrent neural network contains feedback connections [37]. Recurrent neural networks are sensitive and can be adapted to the past inputs [38,39]. The recurrent neural network may be effective in this BDTI forecasting research, however, as suggested by Pearlmutter [40], it is more sensible to begin with trying multi-layered feedforward neural networks to solve the problem before applying recurrent neural network [40].…”

Section: A Feedforwardmentioning

confidence: 99%

Forecasting Baltic Dirty Tanker Index by Applying Wavelet Neural Networks

Fan¹,

Ji²,

Wilmsmeier³

et al. 2013

JTTs

View full text Add to dashboard Cite

Baltic Exchange Dirty Tanker Index (BDTI) is an important assessment index in world dirty tanker shipping industry. Actors in the industry sector can gain numerous benefits from accurate forecasting of the BDTI. However, limitations exist in traditional stochastic and econometric explanation modeling techniques used in freight rate forecasting. At the same time research in shipping index forecasting e.g. BDTI applying artificial intelligent techniques is scarce. This analyses the possibilities to forecast the BDTI by applying Wavelet Neural Networks (WNN). Firstly, the characteristics of traditional and artificial intelligent forecasting techniques are discussed and rationales for choosing WNN are explained. Secondly, the components and features of BDTI will be explicated. After that, the authors delve the determinants and influencing factors behind fluctuations of the BDTI in order to set inputs for WNN forecasting model. The paper examines non-linearity and non-stationary features of the BDTI and elaborates WNN model building procedures. Finally, the comparison of forecasting performance between WNN and ARIMA time series models show that WNN has better forecasting accuracy than traditionally used modeling techniques.

show abstract

“…Dematos et al . () compared performances of feed‐forward and recurrent ANNs against the autoregressive integrated moving average (ARIMA) model for predicting Japanese yen/USD exchange rate, and noted higher accuracy for the ANN models. Wesso () compared the prediction performance of multiple linear regression (MLR), variable parameter regression (VPR) and ANNs.…”

Section: Foreign Exchange Rate Predictionmentioning

confidence: 99%

“…Poddig and Rehkugler (1996), for instance, viewed the stock, bond and currency markets as parts of an integrated finance entity and attempted to capture the implicit nonlinear relationship through an ANN. Dematos et al (1996) compared performances of feed-forward and recurrent ANNs against the autoregressive integrated moving average (ARIMA) model for predicting Japanese yen/USD exchange rate, and noted higher accuracy for the ANN models. Wesso (1996) compared the prediction performance of multiple linear regression (MLR), variable parameter regression (VPR) and ANNs.…”

Section: Introductionmentioning

confidence: 99%

Linear Relationship Between the Aud/Usd Exchange Rate and the Respective Stock Market Indices: A Computational Finance Perspective

Imam

Tickle

Ahmed

et al. 2012

Intell. Sys. Acc. Fin. Mgmt.

View full text Add to dashboard Cite

SUMMARY In the recent era, computational intelligence techniques have found an increased popularity in addressing varied financial issues, including foreign exchange rate prediction. This article, through an intelligent system research framework, relates the Australian dollar (AUD)/US dollar (USD) exchange rate to the Australian and the US stock market indices. Information for exchange rate, All Ordinaries Index (AOI) and Dow Jones Industrial Average (DJI) for the trading days over the period January 1991–May 2011 is considered in this research. Utilizing a set of statistical and computational intelligence techniques, the research establishes that the AUD/USD exchange rate is best estimated by a linear forecast model compared with the nonlinear and ensemble‐based intelligent system models. This research further highlights that, among the competing linear models, the model with both the stock market indices and historical exchange rate values as the predictors is the best forecaster. Parameters of the linear model are deduced through a Monte Carlo stochastic approach. Relative importance of the predictors is also studied, and the influence of historical exchange rates, the immediate impact of AOI and the lagged effect of DJI are noted. Copyright © 2012 John Wiley & Sons, Ltd.

show abstract

Feedforward versus recurrent neural networks for forecasting monthly japanese yen exchange rates

Cited by 28 publications

References 11 publications

Understanding and using time series analyses in addiction research

Understanding and using time series analyses in addiction research

Forecasting Baltic Dirty Tanker Index by Applying Wavelet Neural Networks

Linear Relationship Between the Aud/Usd Exchange Rate and the Respective Stock Market Indices: A Computational Finance Perspective

Contact Info

Product

Resources

About