Data-Based Estimation of Parameters for Time-Inhomogeneous CTMC Birth-and-Death of Call Centers
Data
2023-10-24
Autorzy
TytuĆ czasopisma
ISSN czasopisma
TytuĆ tomu
Wydawca
EUROSIS-ETI
Abstrakt
In this paper, we present a novel approach to estimate parameters of a queuing model using real call center data. Our approach leverages the transient solution of an inhomogeneous continuous-time Markov chain (CTMC) queuing model emulating the behavior of the genuine system. Specifically, to assess the fidelity of our modeling approach, we use an example of approximating the real call center system with an inhomogeneous M(n)/M/c/K + M CTMC model. To do this, we utilize authentic call center data for replicating the behavior of the real system through our model. In particular, the model incorporates the true time-dependent rates for the arrival process, service rates, and the number of available servers. Our analysis focuses on assessing the accuracy of the Markovian assumptions made for modeling customer abandonment during waiting periods. Furthermore, we investigate the performance of our model under two distinct scenarios: overloaded systems and systems operating in a quality-driven mode. By examining these cases, we ascertain the effectiveness of our assumptions in accurately representing the behavior of the call center. Finally, we demonstrate the practical application of our findings by showcasing how a simple and computationally efficient M(n)/M/c/K+M Markovian approximation of a real call center can be used for accurate personnel planning while adhering to service-level constraints.
Opis
SĆowa kluczowe
abandonment, balking, call center, inhomogeneous CTMC, uniformization
Cytowanie
Burak M. (2023). Data-based estimation of parameters for time-inhomogeneous ctmc birth-and-death models of call centers. [In]: Modelling and Simulation 2023 - European Simulation and Modelling Conference 2023, ESM 2023, Toulouse, 24-26 October 2023. pp. 67â75. ISBN 978-9-492-859-28-0. https://hdl.handle.net/20.500.12539/1988