Unpredictable analytics

Recent research from MIT (1) focusing on optimization of path planning for automated underwater vehicles demonstrates that complex engineering problems are solved a lot faster than similar issues in other areas. This is because engineers have been dealing with complex problems in prediction, estimation, control and optimization for many decades. This knowledge, however, has not transferred to more mundane problems in business. For example, in the hottest area of predictive analytics, practicality has taken a back-seat with a focus on complex mathematical techniques at one end and data aggregation and reporting on another.

Better solutions to complex problems use data as an aid and not as a controlling factor. For engineers, the goals are better defined and hence the processes are less cluttered. Those in business, who do not spend sufficient effort to fully define the goals, often get lost in data and lately, the analysis of that data. With hardware vendors peddling faster computers and bigger storage bins, the push has always been on the quantity and not on quality. Equally important are the large software houses and consulting companies focusing on delivering complex closed systems to enterprises, on the hope that such implementations will result in revenue streams to perpetuity.  Both of these effects have resulted in many dead ends and blind alleys for businesses.

Data and analysis are unlikely to improve businesses unless they are able to robustly connect analytics with business value. Without practical approaches to real problem solving, technology and analysis bring little value to businesses.

(1) MIT research: Sometimes the quickest path is not a straight line.Published: Thursday, March 8, 2012 - 14:37 in Mathematics & Economics