Scalable Analysis for Customer Contact Centers

这笔资金用于开发有助于分析客户联系中心绩效的工具。客户联系中心是指通过聊天箱、电子邮件、电话或其他沟通渠道向客户提供服务的组织或实体。这样的中心在社会中发挥着越来越重要的作用。客户联系中心可以是任何规模的，并出现在各种地方。大多数美国公司都雇佣了中小型呼叫中心。对于为客户联系中心建模的多服务器系统，PI建议为各种性能指标(如遇到延迟的客户比例)开发通用、可扩展的近似值。对于小型和大型系统，以及可能不一定负载严重的系统(想想紧急情况下的呼叫中心)，这些近似值在计算上都是高效的。PI将特别关注现实的模型特征，如一般服务时间分布和客户遗弃，这导致开发高维扩散过程的算法的挑战。这项研究将涉及偏微分方程组的数值解、扩散过程、极值理论、大偏差和分析技术等广泛的技术，其中许多技术在这一领域是新的。如果成功，这项研究将导致适用于各种环境的呼叫中心人员配置的单一通用方法。该算法将被实现。提出的软件实现提供了一个实用的工具，使人们能够研究服务时间变化性和耐心时间分布对系统性能的影响。PI旨在使学术研究人员、行业从业者和大学生可以使用该软件。

This grant provides funding for the development of tools that help analyze the performance of customer contact centers. Customer Contact Centers are organizations or entities providing service to customers via chat-box, e-mail, phone, or other communication channels. Such centers play an increasingly dominant role in society. Customer contact centers can be of any size and appear in a variety of places. Most US companies employ moderate- or small-size call centers. For multiserver systems that model customer contact centers, the PIs propose to develop universal, scalable approximations for various performance measures such as the fraction of customers that encounter a delay. These approximations will be computationally efficient for both small and large systems, and for systems that may not necessarily be critically loaded (think of call centers for emergencies). The PIs will pay special attention to realistic model features like general service time distribution and customer abandonment, leading to the challenge of developing algorithms for high dimensional diffusion processes. The proposed research will involve a wide range of techniques like numerical solution of partial differential equations, diffusion processes, extreme value theory, large deviations and analytic techniques; many of which are new in this area.If successful, this research will result in single universal method for call center staffing that is applicable in a variety of settings. The algorithm will be implemented. The proposed software implementation provides a practical tool to allow one to study the impact of service time variability and patience time distribution on system performance. The PIs intend to make the software available to academic researchers, practitioners from industry, and university students.