Practical statistical methods for call centres with a case study addressing urgent medical care delivery

Wooff, D.A.; Stirling, S.G.

Abstract

Our interest is in forecasting for call centres, and in particular out-of-hours call centres (OOHCC) which deal with patient requests for medical advice outside normal working hours. Planning needs accurate forecasts of incoming call volumes. These vary by hour, day, and season, and must account for calendar effects such as Christmas. Using historical data, we explain how to use simple regression models to forecast call volumes arriving on specified days, taking into account calendar effects. We then show how we forecast the pattern of arrivals of calls during a specified day. These result in predictions for volumes of calls arriving for each day of the year, and their pattern of arrival during the day. We show how simulation models may then be used for resource allocation, uncertainty analysis, and staff scheduling. The data are details of call numbers and queue lengths from all parts of the patient-advice process for around five years, for a call centre based in Newcastle-upon-Tyne. There are around 350,000 complete cases in total. The methods are easily extended to other kinds of call centre. We describe the impact Swine flu had on call volumes in the summer of 2009, and our reactions to amend models in order to maintain forecast quality.

Citation

Wooff, D., & Stirling, S. (2015). Practical statistical methods for call centres with a case study addressing urgent medical care delivery. Annals of Operations Research, 233(1), 501-515. https://doi.org/10.1007/s10479-014-1529-2

• Journal Article Type: Article
• Online Publication Date: Jan 16, 2014
• Publication Date: Oct 1, 2015
• Deposit Date: Jun 14, 2012
• Publicly Available Date: Oct 29, 2013
• Journal: Annals of Operations Research
• Print ISSN: 0254-5330
• Electronic ISSN: 1572-9338
• Publisher: Springer
• Peer Reviewed: Peer Reviewed
• Volume: 233
• Issue: 1
• Pages: 501-515
• DOI: https://doi.org/10.1007/s10479-014-1529-2
• Keywords: Call-centre forecasting, Prediction interval, Daily arrival pattern, Nonhomogenous Poisson process, Patient queue, Nurse scheduling.
• Public URL: https://durham-repository.worktribe.com/output/1475962