UCL School of Management

Research project

Data analytics in service systems


Accurately forecasting future demand is crucial for the effective management of any service system. In call centers, managers need reliable ways of forecasting future calls in order to make appropriate operational decisions. This is a hard task since there are usually tens of different call types corresponding to different skills or languages. In this project, we investigate whether the arrival streams of different incoming call types are dependent in practice. Through analysis of real-life data, we propose models which exploit such dependencies and show that our models yield much more accurate forecasts of future demand. 


The findings of this project were useful to the telecom company from which our data was gathered. The new models that we proposed explored new features in the data which were not previousely considered by the forecasting team of that company. Also, the improvement in forecasting accuracy should enable more effective decision-making for its managers. From a theoretical standpoint, the new stochastic models that we propose advance the theory of modelling arrival processes in service systems. Our models are applicable beyond the context of call centers, and are useful in different service settings as well.   

Selected publications

Ibrahim, R., R├ęgnard, N., L'Ecuyer, P., & Shen, H. (2012). On the modeling and forecasting of call center arrivals. Winter Simulation Conference, 23:1. WSC. [link]
Ibrahim, R., & L'Ecuyer, P. (2013). Forecasting call center arrivals: Fixed-effects, mixed-effects, and bivariate models. Manufacturing and Service Operations Management, 15 (1), 72-85. doi:10.1287/msom.1120.0405 [link]

Link to the publication’s UCL Discovery page

Last updated Thursday, 31 July 2014


Research groups

Operations & Technology

Research areas

Management science

Research topics

Applied probability; Applied statistics; Applied stochastic modelling