Session Chair and Organizer, Lingzhou Xue, NISS Assistant Director, opened the NISS Sponsored session entitled, “Big Data Statistical Challenges and Opportunities in Industry” by welcoming speakers and attendees. The session certainly generated much interest. Attendees stood in the back of the hall three deep, while many hardy souls took seats on the floor down the aisle!
Four invited speakers discussed emerging statistical challenges and opportunities in several major industries: insurance, banking, information technology, internet, and pharmaceutical industry.
The first speaker was Dr. Siddhartha (Sid) Dalal, whose talk was titled, “Deep Analytics for Risk Analysis & Mitigation: From NLP, Computer Vision to Sensors.” Sid is a Professor of Professional Practice at Columbia University. Before joining Columbia, he was Chief Data Scientist and Senior Vice President at American International Group and the CTO at RAND Corporation. He described a broad array of Big Data examples and challenges in assessing risks from his experiences in AIG and RAND, which can be addressed by AI and statistical learning methods.
The second speaker was Dr. Jie Chen, who, with Dr. Vijayan (Vijay) Nair, talked about “Machine Learning Techniques for Risk Analysis in Banking.” At Wells Fargo & Company, Vijay is Head of the Statistical Learning and Advanced Computing Group, and Jie is Managing Director in Corporate Model Risk. Jie presented the applications of statistics and machine learning for managing risks in consumer or retail banking, wholesale banking, investment banking and wealth management.
Jie was followed by Dr. Christopher Holloman, the Chief Data Scientist at Information Control Company. who commented on “Sound Statistical Inference from Big Data in the Insurance Industry.” Chris pointed out that the use of new high-resolution data for individualized pricing is one of the most exciting areas for innovation in insurance. In particular, he showed how automobiles containing sensors provide a rich source of information about driver behavior.
The final speaker was Dr. Andrew Smith from Google. In his talk titled “Feature Engineering From Scratch” Andrew emphasized the fundamental importance of feature engineering to the application of statistics and machine learning and discussed its current challenges. The challenge now was to find new ways to uncover important features, what statisticians usually call variables, among the numerous possibilities in big data sets.
James Rosenberger, NISS Director, served as discussant for this popular session, then helped field questions for the speakers from the floor before the session ended.