5 Predictive Analytics Use Cases
December 7, 2012 11:51 AM
Predictive analytics, that long-existent forward-looking cousin of traditional business analytics, has been getting more attention of late because of innovative advances. Lets step back from the hype and check out a few industries and examples of predictive analytics working in the real world. Here are five business use cases of predictive analytics from our archives and trusted experts.
Visions of Risk
Every business wants to peer into the future, but few industries rely so heavily on forecasts as insurance. So it may be of little surprise that insurers have deemed predictive models as a top investment priority. At Utica National Insurance Group, sometimes the predictive models theyve put forth are low on effort but high on unexpected returns. For example, continuously incoming credit reports enable a model that can assess risk appetite based on a range of existing data rather than credit score alone.
Powering Ongoing Performance
Utilities have not always had wide-reaching success with BI implementations. Still, when it comes to predictive analytics, the industry is moving aggressively into the space by 2014, particularly with tools for monitoring usage and outages, according to industry watcher BRIDGE Energy Group. For instance, CenterPoint Energy can now outline analyze automated data from its metering systems to move toward information sharing across new departments and with users.
Steadying Volatile Markets
Recently at the IE Predictive Analytics Summit, data scientist officers at Sears Holding Corporation, the parent company of the retailer, presented the makeup of the more than 100 performance metrics they tap into for predictive financial models. As analyst/blogger Steve Miller wrote about the presentation: The DS group then creates segments using techniques like k-mean clustering, and examines KPI trends against features volatility and momentum that are so important in financial services. With support vector machines and neural nets as their essential classification engines, the authors note off-the-charts efficiency improvements from the efforts.
Finding the Right Person for the Job
Dean Abbot, president of Abbot Analytics, detailed his experience consulting the U.S. Special Forces on data models to help assess new candidates. The models there could be transferred to a predictive framework for any employer looking for the best fit for an open position. Here was Abbots approach to the Special Forces gig and others: The question is, How much does each factor matter? You can find acceptable trade-offs (read: trainable) with a model that gives pertinent weight to questions more vital to intelligence and learning (or whatever is most vital for the position you're looking to fill). And a good target variable is the people who have stayed and succeeded (though that can take a while and not always give a huge pool of results), and a small, quick pool to avoid too much noise.
Location, Location, Location
Much like items trend on the Web, predictive models tied to location technology have sprouted up to gauge experiences and expected outcomes for future events. Foursquare, Facebook and other social network purveyors have implemented analytics along incoming location data to assess upcoming events, and vendors like Pitney Bowes are attaching location-based analytic forecasting with its BI suites.