Large Scale Model Migration Delivers USD 10MM in Savings

Business Objective

Our client, a Fortune 100 company, is among the world’s largest provider of life insurance, P&C insurance, and annuities products serving 90+ million customers. The client wanted to embrace open source technology across the organization and reduce dependency on proprietary tools and technologies. As part of this endeavor, our client wanted to:

— Migrate Analytics models from a license-based analytics tool to open source technology
— Design an end-to-end open source integration strategy
— Design a continuous model monitoring and maintenance framework

Challenges

75% of the analytics use cases lacked documentation around business requirements, data, modeling approach, model evaluation, and business impact determination.
No set process for model development or templates existed for any deliverable and no testing standard applied either
Lack of stable governance as the existing models were not version controlled. There is no audit review process to validate model methodology and results

Solution Methodology

Implemented an end-to-end analytics lifecycle process for large scale model migration ensuring consistency throughout the organization.

Understood analytics use case objectives and its implications in the business process
Identified technical and business stakeholders and connected with them to align on objectives
Gained detailed understanding of the data sources and open source infrastructure set up and dependencies
Designed the governance structure for the model development life cycle. Designed and created standard templates for each stage to main consistency across process
Identified components and methods for process efficiency
Migrated models from traditional (SPSS, SAS) to open source platform (Python) with elaborate testing so that the existing business cycle (upward or downward) is not affected by the change
Created detailed documentation and recommendation for model enhancement and improvement.

Business Impact

USD 10MM estimated savings due to licensing cost, model improvements and process efficiency
~300+ analytics models migrated to open source across 12 Lines of Business
Improved process efficiency with reduced turnaround time and streamlined analytics process by developing reusable components.

Privacy Notice Cookie Policy