Service-Request Distributions in Business Processes and Their Control-Oriented Applications

• Activity Number: 137
• Type: Contributed
• Date/Time: Monday, August 1, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Section on Quality and Productivity
• Author(s): Genady Grabarnik and Yefim Haim Michlin*+ and Laura Shwartz
• Companies: St. John's University and Israel Institute of Technology and IBM Thomas J. Watson Research Center
• Keywords: service ; request process ; sequential test ; comparison test

Abstract

In this paper we analyze statistical characteristics of service requests for several different real business processes, with a view to identifying the laws which govern frequently-occurring distributions of the requests. We argue that the high uncertainty in the service processes necessitates a measuring framework with low dependency on variance. These distributions and their characteristics serve as a base for modeling, optimization and control of the processes. The paper describes how to design comparison tests based on Wald's sequential analysis, for estimating the value and evaluating the deviation of the main measurements of a business process: yield, processing time, throughput and error rate. The obtained characteristics are applied in the design of sequential tests for decision support, effectiveness of modifications, and such updates as resource re-allocation, training or retrai