Predictive analytics lets insurance companies get more granular, analyzing an individual’s account data to determine risk, said Richard Boire, partner at Boire Filler Group, in a phone interview earlier this week.
“What we’ll do is we’ll take [an individual policyholder’s account] and we’ll look at the history,” Boire told us. With predictive analytics, insurance companies can mine historical data on policyholders, looking back 12 months for additional claims or other changes that might result in a different overall risk score.
In the case of property insurance, for example, including riders for building additions or other property alterations usually affects overall risk score, Boire said.
Boire Filler, in Pickering, Ontario, just east of Toronto, builds predictive analytics models for Canadian insurance companies. Its clients can then set individual policy prices based on changing risk factors.
For example, a female driver in her 20s would have a lower risk score and lower insurance rate than a male of the same age. However, the same female driver with three new claims on her auto insurance over the last 12 months would see her risk score climb significantly.
Conversely, a 20-year-old male who would normally see a high risk score based on age and gender might see that drop with a long enough driving history free from any claims. Heck, if this guy also wants property insurance his good driving may earn him points there, too. See this video blog on how one insurance company apparently has found a connection between lousy driving and careless home ownership.
Boire will join us Friday, March 2, for a live e-chat on predictive analytics in insurance pricing at 2 p.m. EST here on AllAnalytics.com. He will answer questions about how to best use data mining tools to detect risk scores for individual policyholders. If you're attending next week's Predictive Analytics World conference, in San Francisco, you can catch him there, too. Boire will be presenting on Monday, March 5.
How might the insurance industry benefit from predictive analytics in your view? Leave your thoughts on the board below, and be sure to join us for the live chat on Friday and feel free to drop some questions in the chat interface before the event to get us started.