Innovative Statistical Applications to Business

• Activity Number: 61
• Type: Topic Contributed
• Date/Time: Sunday, July 29, 2012 : 4:00 PM to 5:50 PM
• Sponsor: Business and Economic Statistics Section
• Author(s): Han Ye*+
• Companies: The University of North Carolina at Chapel Hill
• Keywords: forecasting ; multiple-stream arrivals ; vector time series ; dimension reduction ; chance-constraint ; integer programming

Abstract

Forecasting call arrivals plays a crucial role in call center staffing and scheduling process. Most call center statistical forecasting methods focus on a single stream. We incorporate the dependence between different streams and develop methods to simultaneously forecast multiple-stream arrivals in a call center. We first reduce the dimensionality of the arrival data matrix using regression techniques and apply the vector time series model to obtain the forecasts. We then implement the chance-constraint staffing algorithm to assess the operational effect of our forecasts. Experiment on a real call center data set shows that the multiple-stream approach provides more accurate forecasts and the consequential staffing policy meets the quality of service objective better, compared with the single-stream approach. Simulation studies indicate the forecasting and operating benefits of multiple-stream approach vary by type and strength of the dependence among streams.