One of our favourite quotes comes from the Scottish poet, Andrew Lang: “An unsophisticated forecaster uses statistics as a drunken man uses lamp-posts – for support rather than for illumination.” This seems to summarize a problem with current systems modernization and analytic initiatives.
Numbers geeks (present company included) are attracted to titles such as the following which recently appeared in PropertyCasualty360:
Building Predictive Models that Estimate Differentials for P&C Claim Risk
As good as it was, it was the opening 2 sentences of the article that took our breath away:
“Building predictive models to estimate claim risk is not revolutionary. We all know that the techniques of using multivariate analysis and regression to predict a given outcome or behavior have been around since the end of World War II. ”
Actually, not everyone knows that. The problem is that many people who don’t know, act as if they do, and make decisions on business data analytic systems that would best be left to others. Rachel Alt-Simmons, from SAS, points this out in a recent article in Information&Technology She writes: “Over the past decade, insurers have made significant investments in their operational systems, but very few have used that opportunity to capitalize on the new wealth of information or capabilities that those systems enable.”
Alt-Simmons notes that modern policy and claims administrations systems frequently include data warehouse and analytic components at the outset which are later scrapped or deferred to keep the overall project on scope. The assumption is that these can later be fitted in. Unfortunately, as Alt-Simmons notes: “post-implementation, the data isn’t in a format that’s easily consumed, or needed data elements may not be persistent in the system,” rendering the analytical components infective or useless without costly system changes.
About 10 years ago, courses started to emerge with titles such as ‘Financial management for non-financial managers.” The concept was to provide sufficient skills for underwriting, marketing, and event IT managers to be able to read and understand corporate financial statements. Perhaps this concept could be extended to analytic concepts. Here’s a suggested title: “Regression analysis for non-regressing managers.” Well, maybe that needs work ….