I chose to work in SPSS Modeler over other predictive modeling tools largely because its visual process flow and additional features allow non-programmers and predictive modelers to go from a blank page to a finished model more quickly than other tools.
As a long-time certified Enterprise Optimizer (EO) developer, my next priority was to ask IBM where prescriptive optimization modeling fit in with their offerings, and how they leverage the CPLEX integer optimization engine that they own (which is purchased from iLog and optionally available in EO).
Over the course of the conference, predictive modeling ruled the day. SPSS models were at the heart of the many ways users scale and automate use.
After searching a show floor, I was finally able to find someone who knew what CPLEX was, and was able to show me how it interacted with IBM’s set of products. In the demonstration (interacting with IBM Analytical Decision Management), CPLEX was used with the results of a predictive model, as well as a set of simple constraints, to optimize a set of decisions.
While their interface and approach simplified the use of the modeling tool, the model was also very simply bounded by the results of the predictive model. Increased complexity required users to fall back to the programming environment for OPL.
The true challenge of creating a flexible, non-programming, prescriptive modeling interface has not been directly addressed by IBM.
So, while the CPLEX engine is in use by IBM for prescriptive modeling, the true challenge of creating a flexible non-programmer interface (to create models and leverage CPLEX) has not been directly addressed by IBM.
Shocked? Intrigued? Dive deeper into why IBM's CPLEX Optimizer isn't all that it claims to be in the product comparison below.
