SIGLEAvailable from INIST (FR), Document Supply Service, under shelf-number : DO 6748 / INIST-CNR... more SIGLEAvailable from INIST (FR), Document Supply Service, under shelf-number : DO 6748 / INIST-CNRS - Institut de l'Information Scientifique et TechniqueFRFranc
A working paper in the 1NSEAD Working Paper Series is intended as a means whereby a faculty resea... more A working paper in the 1NSEAD Working Paper Series is intended as a means whereby a faculty researcher's thoughts and findings may be communicated to interested readers. The paper should be considered preliminary in nature and may require revision. Printed at INSEAD, Fontainebleau, France.
SIGLEAvailable at INIST (FR), Document Supply Service, under shelf-number : DO 997 / INIST-CNRS -... more SIGLEAvailable at INIST (FR), Document Supply Service, under shelf-number : DO 997 / INIST-CNRS - Institut de l'Information Scientifique et TechniqueFRFranc
SIGLEAvailable at INIST (FR), Document Supply Service, under shelf-number : DO 3434 / INIST-CNRS ... more SIGLEAvailable at INIST (FR), Document Supply Service, under shelf-number : DO 3434 / INIST-CNRS - Institut de l'Information Scientifique et TechniqueFRFranc
During 2007 we conducted an empirical evaluation of the accuracy of artificial Neural Networks (N... more During 2007 we conducted an empirical evaluation of the accuracy of artificial Neural Networks (NN) and other methods of Computational Intelligence (CI) in time series prediction through a dedicated forecasting competition: the NN3 (www.neural-forecasting-competition.com). The competition aimed to resolve two research questions: (a) what is the current performance of CI methods in comparison to established statistical forecasting methods, and (b)
The major aims of the M3-Competition are to extend and replicate the findings of the M-and M2-one... more The major aims of the M3-Competition are to extend and replicate the findings of the M-and M2-ones. The extension involves the inclusion of more researchers, more methods (in particular in the area of neural networks and expert systems) and most importantly more series as the database of the M3-Competition has been enlarged to include 3003 time series. In terms of replication our purpose is to determine if the four major conclusions of the M-and M2-Competitions:
The canton of Valais has 19 different destinations. At the lowest level, tourism frequentation da... more The canton of Valais has 19 different destinations. At the lowest level, tourism frequentation data , namely overnights, are collected for each of its 127 municipalities. The destinations are classified into three majors regions and the overnights sorted into Swiss residents and foreigners. It this way, the hierarchy of time series is composed of a tree of depth equal to 5. When forecasting the total tourism frequentation for the Valais at the aggregate level, the effect of winter holidays is difficult to model in the forecast, i.e., intervention variables. On the one hand, the 26 cantons set school vacation independently but try to stagger them as much as possible. On the other hand, tourists from different cantons have different preferred Valais ski resorts but these preferences are only partially known. The aim of this research is to assess whether the use of regional grouped time series forecast give better forecasts than on the aggregates levels. The ultimate objective of the c...
Journal of the Royal Statistical Society. Series A (General), 1979
ln the last few decades matiy methods have become available for forecasting. As always, when alte... more ln the last few decades matiy methods have become available for forecasting. As always, when alternatives exist, choices need to be made so that an appropriate forecasting method can be selected and used for the specific situation being considered. This paper reports the results of a forecasting competition that provides information to facilitate such choice. Seven experts in each of the 24 methods forecasted up to 1001 series for six up to eighteen time horizons. The results of the competition are presented in this paper whose purpose is to provide empirical evidence about differences found to exist among the various extrapolative (time series) methods used in the competition.
The purpose of this paper is to apply the Box±Jenkins methodology to ARIMA models and determine t... more The purpose of this paper is to apply the Box±Jenkins methodology to ARIMA models and determine the reasons why in empirical tests it is found that the post-sample forecasting the accuracy of such models is generally worse than much simpler time series methods. The paper concludes that the major problem is the way of making the series stationary in its mean (i.e. the method of dierencing) that has been proposed by Box and Jenkins. If alternative approaches are utilized to remove and extrapolate the trend in the data, ARMA models outperform the models selected through Box±Jenkins methodology. In addition, it is shown that using ARMA models to seasonally adjusted data slightly improves post-sample accuracies while simplifying the use of ARMA models. It is also con®rmed that transformations slightly improve post-sample forecasting accuracy, particularly for long forecasting horizons. Finally, it is demonstrated that AR(1), AR(2) and ARMA(1,1) models can produce more accurate post-sample forecasts than those found through the application of Box± Jenkins methodology. # 1997 by John Wiley & Sons, Ltd.
SIGLEAvailable from INIST (FR), Document Supply Service, under shelf-number : DO 6748 / INIST-CNR... more SIGLEAvailable from INIST (FR), Document Supply Service, under shelf-number : DO 6748 / INIST-CNRS - Institut de l'Information Scientifique et TechniqueFRFranc
A working paper in the 1NSEAD Working Paper Series is intended as a means whereby a faculty resea... more A working paper in the 1NSEAD Working Paper Series is intended as a means whereby a faculty researcher's thoughts and findings may be communicated to interested readers. The paper should be considered preliminary in nature and may require revision. Printed at INSEAD, Fontainebleau, France.
SIGLEAvailable at INIST (FR), Document Supply Service, under shelf-number : DO 997 / INIST-CNRS -... more SIGLEAvailable at INIST (FR), Document Supply Service, under shelf-number : DO 997 / INIST-CNRS - Institut de l'Information Scientifique et TechniqueFRFranc
SIGLEAvailable at INIST (FR), Document Supply Service, under shelf-number : DO 3434 / INIST-CNRS ... more SIGLEAvailable at INIST (FR), Document Supply Service, under shelf-number : DO 3434 / INIST-CNRS - Institut de l'Information Scientifique et TechniqueFRFranc
During 2007 we conducted an empirical evaluation of the accuracy of artificial Neural Networks (N... more During 2007 we conducted an empirical evaluation of the accuracy of artificial Neural Networks (NN) and other methods of Computational Intelligence (CI) in time series prediction through a dedicated forecasting competition: the NN3 (www.neural-forecasting-competition.com). The competition aimed to resolve two research questions: (a) what is the current performance of CI methods in comparison to established statistical forecasting methods, and (b)
The major aims of the M3-Competition are to extend and replicate the findings of the M-and M2-one... more The major aims of the M3-Competition are to extend and replicate the findings of the M-and M2-ones. The extension involves the inclusion of more researchers, more methods (in particular in the area of neural networks and expert systems) and most importantly more series as the database of the M3-Competition has been enlarged to include 3003 time series. In terms of replication our purpose is to determine if the four major conclusions of the M-and M2-Competitions:
The canton of Valais has 19 different destinations. At the lowest level, tourism frequentation da... more The canton of Valais has 19 different destinations. At the lowest level, tourism frequentation data , namely overnights, are collected for each of its 127 municipalities. The destinations are classified into three majors regions and the overnights sorted into Swiss residents and foreigners. It this way, the hierarchy of time series is composed of a tree of depth equal to 5. When forecasting the total tourism frequentation for the Valais at the aggregate level, the effect of winter holidays is difficult to model in the forecast, i.e., intervention variables. On the one hand, the 26 cantons set school vacation independently but try to stagger them as much as possible. On the other hand, tourists from different cantons have different preferred Valais ski resorts but these preferences are only partially known. The aim of this research is to assess whether the use of regional grouped time series forecast give better forecasts than on the aggregates levels. The ultimate objective of the c...
Journal of the Royal Statistical Society. Series A (General), 1979
ln the last few decades matiy methods have become available for forecasting. As always, when alte... more ln the last few decades matiy methods have become available for forecasting. As always, when alternatives exist, choices need to be made so that an appropriate forecasting method can be selected and used for the specific situation being considered. This paper reports the results of a forecasting competition that provides information to facilitate such choice. Seven experts in each of the 24 methods forecasted up to 1001 series for six up to eighteen time horizons. The results of the competition are presented in this paper whose purpose is to provide empirical evidence about differences found to exist among the various extrapolative (time series) methods used in the competition.
The purpose of this paper is to apply the Box±Jenkins methodology to ARIMA models and determine t... more The purpose of this paper is to apply the Box±Jenkins methodology to ARIMA models and determine the reasons why in empirical tests it is found that the post-sample forecasting the accuracy of such models is generally worse than much simpler time series methods. The paper concludes that the major problem is the way of making the series stationary in its mean (i.e. the method of dierencing) that has been proposed by Box and Jenkins. If alternative approaches are utilized to remove and extrapolate the trend in the data, ARMA models outperform the models selected through Box±Jenkins methodology. In addition, it is shown that using ARMA models to seasonally adjusted data slightly improves post-sample accuracies while simplifying the use of ARMA models. It is also con®rmed that transformations slightly improve post-sample forecasting accuracy, particularly for long forecasting horizons. Finally, it is demonstrated that AR(1), AR(2) and ARMA(1,1) models can produce more accurate post-sample forecasts than those found through the application of Box± Jenkins methodology. # 1997 by John Wiley & Sons, Ltd.
Uploads
Papers by Michele Hibon