Last summer, I went online to purchase a new inner tube for my bicycle tire. It was a one-time requirement that was fixed and done with. But then I noticed for weeks afterwards, I was being served online ads for bicycle tire inner tubes when surfing the web. The algorithm that picked up my data trail on shopping for inner tubes had predicted that I would have an ongoing, insatiable need for bicycle inner tubes! (I should have told it about my plans to try to avoid sharp objects laying on pavements.)
Such are the pitfalls with algorithms they can pick up some nuggets of data, and draw erroneous conclusions. I was thinking about this when INN colleague Chris McMahon recently posted the results of an intriguing survey, disclosing how underwriters are resistant to predictive analytics out of fears it will replace their jobs. While 45 percent report they are incorporating predictive analytics solutions into their companies, 24 percent of underwriters believe that their experience is more valuable than a predictive score when assessing risk, the survey finds. One in four underwriters believe predictive analytics will replace their jobs entirely.
While fears of job obsolescence are entirely justified, such moves to automation also mean opportunities as well. While routine, relatively straightforward policy engagements may breeze through predictive analytics engines tied to underwriting rules, the special cases will still require creative thinking. Plus, there’s a likelihood that many jobs in the future will be elevated to advisory and consultative roles, versus down-in-the-weeds number crunching. And remember, technologies such as predictive analytics are not a slam-dunk, either. The need for organizational collaboration gets even more intense, as does the need for critical thinking. The business needs to decide exactly what kind of data matters, and how this may change day to day.
Gregory Piatetsky, in his KDNuggets blog, observes that predictive analytics is great for high-volume, routine day-to-day decisions, but nothing will replace human creativity. He also points to a post by data scientist Mikio Braun, who makes the point that predictive analytics is hard.
Braun makes four observations about the difficulties and pitfalls about relying too much on machine-generated analytics: “Data analysis is so easy to get wrong; it’s too easy to lie to yourself about it working; it’s very hard to tell whether it could work if it doesn’t; there is no free lunch.”
“’Garbage In Garbage Out’ is even more true for data analysis,” he states. “And there are so many ways to get this wrong, like discarding important information in a preprocessing step, or accidentally working on the wrong variables. The algorithms don’t care, they’ll give you a result anyway.” Even a seemingly well-performing algorithm could be steering things in the wrong direction, he cautions.
Data analytics tools are just that tools. There still needs to be business savvy, skills and critical thinking to ensure that any type of analytics is working for the business, to advance the business and not be turning away opportunities, or feeding the wrong opportunities.
Register or login for access to this item and much more
All Digital Insurance content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access