IBM Watson and Pacific Life

By John Peach | July 31, 2018

These talks were given on July 31, 2018.

This month we great talks from UCI’s first set of graduates in the Master of Science, Business Analytics program. The talks are based on the analysis done by a team of graduate students during their five-month Capstone projects. As an interesting twice, one team did the analysis using python and the other used R. Both are great languages for analytical work.

Talk 1: Pacific Life’s Deferred Annuity

Our most essential task in this project is to review the mortality experience from Pacific Life. In particular, we are looking into the mortality from Pacific Life’s Deferred Annuity block of business. A Deferred Annuity is a retirement investment vehicle and functions like a mutual fund if managed efficiently. Pacific Life is internally required to use their current model to run periodic projections for the purposes of hedging, pricing and various other activities. Our goal is to identify and analyze the policy attributes that led to the disparity from the model and improve the model according to the results of the analysis and potentially provide business intelligence regarding mortality that can support marketing and operational strategies.

This presentation can be found at:

Talk 2: IBM Watson - The True Benefits of HR contracts

Our team project is to help determine the true benefits costs in Human Resource contracts and we were working with IBM and the City of Los Angeles for our capstone. We used IBM Watson software (mainly Knowledge Studio and Natural Language Understanding) to conduct an analysis and our target data was the MOUs published on the City of LA website. Throughout the five-month project, we were given timetables in which we were to essentially learn the City MOUs breakdown and understand the MOUs. From the City’s part, questions/possibilities were brought up as in, what are the common and unique benefits for each MOU? Can Watson help identify more irrelevant information in City contracts, so they can be amended? Can we use the potential model to compare other City employment contracts/MOUs to their own? And lastly, how will the changes of benefit costs bring to social issues such as crime rate? Through creating dictionaries, applying pre-annotators, doing human annotating, training and evaluating the models, we achieved an overall model F1 score of 0.8. For detailed statistics, we got 0.5 in recall for ‘benefits’ and 0.26 for ‘eligibility’. Recalls for other entities are as high as above 0.7. We helped the City of LA in saving contract processing time and the model was able to analyze contract (average 19,000 words) within 10mins.

This presentation can be found at:

Agenda: 6:30-7:00 Welcome and Networking 7:00-8:00 * Introduction * Talk 1: Pacific Life’s Deferred Annuity (Wanyi Huang) * Talk 2: IBM Watson - The True Benefits of HR contracts (Gloria Gong) 8:00-8:30 Clean-up and networking

comments powered by Disqus