PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

Different approaches to the tuning of the RCPSP algorithm RaR
Janez Žerovnik and Igor Pesek
In: the 10th International Symposium on Operational Research SOR '09, September 23-25, 2009, Nova Gorica, Slovenia.


Tuning the algorithms can be very time consuming and not always successful. In this paper we present two different methods for determining the best values for our algorithm RaR which was developed for the well known RCPSP problem. First method is usual “brute force” method, which uses average deviation as main decision tool. Second method uses more sophisticated statistical methods for more reliable parameter tuning. We compare both methods and present the results of our experiments.

EPrint Type:Conference or Workshop Item (Paper)
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Theory & Algorithms
ID Code:6712
Deposited By:Igor Pesek
Deposited On:08 March 2010