NN5 HomeMotivationDatasetsInstructionsFAQ & HelpConferencesPublicationsResultsPast Competitons

NN5 Competition Results

Results on the Complete Dataset of 111 Time Series

The results will be disclosed in mid 2008.

COMPLETE DATASET   Official Rank         Unofficial Ranks    
#ID Contender Name Mean SMAPE NN & CI Methods Statistical Methods All
Methods
Conference
Presentation
Method
Description
B02 Wildi 19,9%   1 1  
C23 Andrawis 20,4% 1   2  
C12 Vogel 20,5% 2   3    
C10 D'yakonov 20,6% 3   4    
B08 Noncheva 21,1%   2 5    
C06 Rauch 21,7% 4   6    
C19 Luna 21,8% 5   7    
B05 Lagoo 21,9%   3 8    
C01 Wichard 22,1% 6   9    
C17 Gao 22,3% 7   10    
C08 Puma-Villanueva 23,7% 8   11    
B01 Autobox (Reilly) 24,1%   4 12    
B04 Lewicke  24,5%   5 13    
B07 Brentnall  24,8%   6 14    
C09 Dang 25,3% 9   15    
C05 Pasero 25,3% 10   16    
C24 Adeodato  25,3% 11   17    
C25 undisclosed 26,8% 12   18    
C20 undisclosed 27,3% 13   19    
C26 Tung 28,1% 14   20    
B12  Nave Seasonal 28,8%   7 21    
C14 undisclosed 33,1% 15   22    
C28 undisclosed 36,3% 16   23    
C22 undisclosed 41,3% 17   24    
C02 undisclosed 45,4% 18   25    
B11  Nave Level 48,4%   8 26    
C21 undisclosed 53,5% 19   27    

 

Results on the Reduced Dataset of 11 Time Series (subset of the complete dataset)

The results will be disclosed in mid 2008.

REDUCED DATASET   Official Rank         Unofficial Ranks    
#ID Contender Name Mean SMAPE NN & CI Methods Stats. Methods All Methods Conference
Presentation
Method
Description
B02 Wildi 17,6%   1 1    
C06 Rauch 19,0% 1   2    
C10 D'yakonov 19,9% 2   3    
C23 Andrawis 20,5% 3   4    
B05 Lagoo 21,0%   2 5    
C19 Luna 21,1% 4   6    
C04 Hung 21,3% 5   7    
B08 Noncheva 21,7%   3 8    
C07 Gutierrez 21,9% 6   9    
C01 Wichard 22,4% 7   10    
C12 Vogel 22,4% 8   11    
C08 Puma-Villanueva 23,1% 9   12    
C17 Gao 23,2% 10   13    
C05 Pasero 23,2% 11   14    
B07 Brentnall  23,4%   4 15    
B09 Merkusheva 23,8%   5 16    
B01 Reilly 23,9%   6 17    
B04 Lewicke  24,5%   7 18    
C20 Teddy 24,7% 12   19  
B03 Beadle 24,9%   8 20    
C18 Fillon 25,4% 13   21    
C09 Dang 25,4% 14   22    
C25 Coyle 25,9% 15   23    
B12  Nave Seasonal 27,8%   9 24    
C29 Rabie 28,2% 16   25    
C24 Adeodato  29,9% 17   26    
C15 undisclosed 30,2% 18   27    
C16 undisclosed 30,6% 19   28    
C11 undisclosed 31,5% 20   29    
C26 Tung 34,6% 21   30    
C13 undisclosed 34,7% 22   31    
C14 undisclosed 35,5% 23   32    
C22 undisclosed 37,6% 24   33    
C03 Carbajal 40,1% 25   34    
C02 undisclosed 40,2% 26   35    
C28 undisclosed 40,5% 27   36    
C27 undisclosed 47,4% 28   37    
B11  Nave Level 48,6%   10 38    
C21 undisclosed 50,3% 29   39    

 

Submissions in RED are statistical methods that entered the competition as "benchmarks". They can either be existing and established statistical methods or novel methods entered to be evaluated through the competition ). Submissions in BLUE are CI methods that entered the competition as "benchmarks" but were computed by the organisers as points of reference. Only original submissions with methods from computational Intelligence are eligible to win the competition (no benchmarks, no statistical methods and were in part calculated by the competition supervisors).

 






 

Important Dates

18 February 2008
Start of the NN5 daily time series forecasting competition
18 May 2008 
Submission deadline for predictions of 11 and 111 time series
1-6 June 2008
NN5 special session at the World Congress on Computational Intelligence (WCCI'08), Hong Kong, China
23-26 June 2008
NN5 special session at the International Symposium on Forecasting (ISF'08), Nice, France
14-17 July 2008
NN5 special session at the International Conference on Data Mining (DMIN'08) Las Vegas, USA

 

 

  2006 BI3S-lab - Hamburg, Germany - All rights reserved - Questions, Comments and Enquiries via eMail -  [Impressum & Disclaimer]
The Knowledge Portal on Forecasting with Neural Networks @ www.neural-forecasting.com -
last update: 18.10.2006