EURO Working Group on Stochastic Modelling

FIRST MEETING: 19-21 APRIL 2006, AMSTERDAM

meeting 1
     program
     practical information
     register
     participants
meeting 2
meeting 3

Abstracts

Sandjai Bhulai, Nearest Neighbour Algorithms for Forecasting Call Arrivals in Call Centers

We study a nearest neighbour algorithm for forecasting call arrivals to call centers. The algorithm does not require an underlying model for the arrival rates and it can be applied to historical data without pre-processing it. We show that this class of algorithms provides a more accurate forecast when compared to the conventional method that simply takes averages. The nearest neighbour algorithm with the Pearson correlation distance function is also able to take correlation structures, that are usually found in call center data, into account. Numerical experiments show that this algorithm provides smaller errors in the forecast and better staffing levels in call centers. The results can be used for a more flexible workforce management in call centers.