Is Your Crystal Ball Foggy? A Practical Framework for Deploying Predictive Models 
Jeff Amspaugh, Process Improvement Lead; Dale Crawford, Measurement Lead; Jeff Bridgman, Measurement Specialist
Accenture

Introduction:

Hosted by Accenture, this presentation will cover best practices in predictive modeling. Whether you've successfully developed and deployed predictive models, or you are embarking upon your first, this workshop will provide practical insights, lessons learned and a tried-and-true framework that you can put to use in your environment.

Learning Objectives:

  • Practical interpretation of Level 4 predictive modeling practices
  • Proven framework for developing, deploying, and institutionalizing predictive models that participants can leverage in their organization
  • Lessons learned from the Accenture Quality & Process Improvement teams predictive model deployment experience

Outline:

  • Company Profile
  • Accenture's Quality Program
  • Accenture CMMI Ò Journey
  • CMMI Ò Level 4 Process Areas
  • Predictive Model Experience
  • Framework for Developing & Deploying Predictive Models
  • Identify
  • Develop
  • Pilot
  • Calibrate and Enhance Model
  • Conclusion

Biography:

Mr. Amspaugh has over 15 years of experience analyzing, designing, and implementing information systems with Federal Government clients, including the US Air Force, US Postal Service, Department of Housing and Urban Development, and DARPA. He has 7+ years of project management experience, managing complex systems integrations and obtained his PMP certification in 2003. He has led functional analysis, design, and test teams and helped develop the technical architecture for custom systems utilizing Microsoft.NET and J2EE technologies. For the last 2.5 years, Mr. Amspaugh has served as the Process Improvement Manager for the Accenture Government North America Quality and Process Improvement (QPI) program and has participated in CMMI ® Level 4 appraisals as an appraiser and participant. The Process Improvement (PI) Manager is accountable for the QPI measurement program, training/communication program and methodology/tools program. This entails all tools, methodologies, communications, training and measures needed to ensure the Accenture Government North America Operating Unit can successfully implement according to Accenture quality policies and maintain a Level 4 CMMI ® rating. This includes the development, piloting, and analysis of process-performance models. Mr. Amspaugh also represents Accenture as a subject matter expert on the CMMI ® quality model and Accenture's Government North America quality initiatives in various external and internal forums.

Mr. Crawford has over 14 years of consulting, system development, and program management experience with Federal Government clients including the Department of Homeland Security, US Postal Service, and the Internal Revenue Service. For the last 7 years, Mr. Crawford has led several program management and performance measurement initiatives for the Accenture Government North America Operating Unit, and is currently the Measurement Lead for the QPI Program. Mr. Crawford participated in Accenture's most recent CMMI ® Level 4 appraisal, and manages the development and deployment of the organization's Level 4 capabilities.

Mr. Bridgman has been at Accenture for five years, working as an analyst on several information systems with Federal Government clients, including the US Postal Service, US Army, and DIA. For the last 2 years, Mr. Bridgman has served as a Measurement Specialist for the Accenture Government North America QPI Program and participated in a CMMI ® Level 4 appraisal. The Measurement Specialist helps in the day-to-day operation of the QPI measurement program, performing the analysis, reporting, liaising, and process improvement duties necessary to maintain a Level 4 CMMI ® rating. Mr. Bridgman serves as a subject matter expert on Accenture metrics and models, and recently led the development and deployment for one of the organization's three predictive process-performance models.