Complex Methods Do Not Always Give Better Forecasts
published: cw 38, 2005 in eLogistics & eFulfillmentSince 1979 different forecasting methods are being tested on reliability and accuracy. This takes place in the so called Makridakis Competitions, or M-Competitions. The main conclusions of the latest M3 competition were published in the latest issue ‘International Journal of Forecasting’ van July/September 2005.
The authors of the article introduced a new methodology that has not previously been used to evaluate economic forecasts: multiple comparisons. They used this technique to compare each method against the best and against the mean, and concluded that the accuracy of the various methods does differ significantly, and that some methods are significantly better than others.
The study confirms that there is no relationship between complexity and accuracy but also show that there is a significant relationship among the various measures of accuracy. Par example the more complex method like ‘Artificial Neural Networks’ does not give per default better results as a less complex method like ‘Single Exponential Smoothing’. The challenge therefore is to determine the method with the best result for your specific need. However there is one method with an higher then average score in most situations. This is the so called Theta-method, a relatively young decomposition technique. In this method are seasonal patterns as well as long and short term trends extrapolated en combined to one forecast. Finally, the authors find that the M3 conclusion that a combination of methods is better than that of the methods being combined was not proven.
Sources: International Journal of Forecasting and MARS Management Science
----- Advertisement -----
Use this powerful tool to expand your professional vocabulary and ensure that everyone on your team is speaking the same language. www.theKnowledgeTransfer.com |
 paperback student version $ 19,99 hardcover executive version $ 29,99 |
print terms
