Combining neural networks and ARIMA models for hourly temperature forecast

Hippert, Henrique S.; Pedreira, Carlos Eduardo; Souza, Reinaldo Castro

doi:10.1109/ijcnn.2000.860807

Cited by 38 publications

(25 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Research in time series forecasting argues that predictive performance improves the combined models. (Bishop (1994), Clemen (1989), Hansen et al (2003), Hibbert et al (2000), Terui et al (2002), Tseng et al (2002), Zhang, (2003, Zhang et al (2005)). …”

Section: Literature Reviewmentioning

confidence: 99%

Modelling and trading the Greek stock market with mixed neural network models

Dunis¹,

Laws

Karathanasopoulos³

2011

Applied Financial Economics

View full text Add to dashboard Cite

In this paper, a mixed methodology that combines both the ARMA and NNR models is proposed to take advantage of the unique strength of ARMA and NNR models in linear and nonlinear modelling. Experimental results with real data sets indicate that the combined model can be an effective way to improve forecasting accuracy achieved by either of the models used separately. The motivation for this paper is to investigate the use of alternative novel neural network architectures when applied to the task of forecasting and trading the ASE 20 Greek Index using only autoregressive terms as inputs. This is done by benchmarking the forecasting performance of six different neural network designs representing a Higher Order Neural Network (HONN), a Recurrent Network (RNN), a classic Multilayer Percepton (MLP), a Mixed Higher Order Neural Network, a Mixed Recurrent Neural Network and a Mixed Multilayer Percepton Neural Network with some traditional techniques, either statistical such as a an autoregressive moving average model (ARMA), or technical such as a moving average convergence/divergence model (MACD), plus a naïve trading strategy. More specifically, the trading performance of all models is investigated in a forecast and trading simulation on ASE 20 fixing time series over the period 2001-2008 using the last one and a half year for out-of-sample testing. We use the ASE 20 daily fixing as many financial institutions are ready to trade at this level and it is therefore possible to leave orders with a bank for business to be transacted on that basis.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

Modelling and trading the Greek stock market with mixed neural network models

Dunis¹,

Laws

Karathanasopoulos³

2011

Applied Financial Economics

View full text Add to dashboard Cite

show abstract

“…Hybrid algorithms combining traditional time series models with neural networks have been shown to produce more accurate forecasting results than traditional statistical processes. Reference [5] proposes a hybrid forecasting method that combines multilayer neural networks with linear models. The linear models are used to forecast hourly temperatures; the results of the linear process are then used as input to the neural network [5].…”

Section: Introductionmentioning

confidence: 99%

Hybrid time series-bayesian neural network short-term load forecasting with a new input selection method

Ghofrani

West

2015

2015 IEEE Power &Amp; Energy Society General Meeting

View full text Add to dashboard Cite

This paper proposes a hybrid short-term load forecasting (STLF) framework with a new, more efficient, input selection method. Correlation analysis and ℓ 2 -norm are used in combination to select suitable inputs to individual Bayesian neural networks (BNNs), which are used to forecast the load. Forecast outputs are then weighted using calculated weighting coefficients and summed to obtain the final forecast for a particular day. New England load data is used to assess the accuracy and performance of the proposed framework; furthermore, a comparison of the proposed STLF with classic time-series methods shows a significant improvement in the accuracy of the load forecast.

show abstract

“…Research in time series forecasting argues that predictive performance improves in combined models. (Bishop (1994), Clemen (1989), Hansen et al (2003), Hibbert et al (2000), Terui et al (2002), Tseng et al (2002), Zhang, (2003, Zhang et al (2005).…”

Section: Literature Reviewmentioning

confidence: 99%

GP algorithm versus hybrid and mixed neural networks

Dunis

Laws

Karathanasopoulos

2013

The European Journal of Finance

View full text Add to dashboard Cite

In the current paper we present an integrated genetic programming environment, called java GP Modelling. The java GP Modelling environment is an implementation of the steady-state genetic programming algorithm. That algorithm evolves tree based structures that represent models of inputoutput relation of a system. The motivation of this paper is to compare the GP algorithm with neural network architectures when applied to the task of forecasting and trading the ASE 20 Greek Index using only autoregressive terms as inputs. This is done by benchmarking the forecasting performance of the GP algorithm and 6 different ARMA-Neural Network combination designs representing a Hybrid, Mixed Higher Order Neural Network (HONN), a Hybrid, Mixed Recurrent Network (RNN), a Hybrid, Mixed classic Multilayer Perceptron (MLP) with some traditional techniques, either statistical such as a an autoregressive moving average model (ARMA), or technical such as a moving average convergence/divergence model (MACD), plus a naïve trading strategy. More specifically, the trading performance of all models is investigated in a forecast and trading simulation on ASE 20 time series closing prices over the period 2001-2008 using the last one and a half years for out-of-sample testing. We use the ASE 20 daily series as many financial institutions are ready to trade at this level and it is therefore possible to leave orders with a bank for business to be transacted on that basis.As it turns out, the GP model does remarkably well and outperforms all other models in a simple trading simulation exercise. This is also the case when more sophisticated trading strategies using confirmation filters and leverage are applied, as the GP model still produces better results and outperforms all other neural network and traditional statistical models in terms of annualised return.

show abstract

Combining neural networks and ARIMA models for hourly temperature forecast

Cited by 38 publications

References 6 publications

Modelling and trading the Greek stock market with mixed neural network models

Modelling and trading the Greek stock market with mixed neural network models

Hybrid time series-bayesian neural network short-term load forecasting with a new input selection method

GP algorithm versus hybrid and mixed neural networks

Contact Info

Product

Resources

About