Prescriptive Analytics Modeling for PythonThere are two kinds of data scientists: those who limit themselves to data analysis, and those who care about turning insights into actions. The latter look for acti Why mathematical optimization? Because it is the only technique that lets you solve a problem by defining an objective, and constraints on how you can reach that objective. Other techniques require you to explicit all steps that can lead to a solution (aka business rules), or require you to build a model of how the system you try to influence behaves (aka simulation). With optimization, you simply need to express the properties you want to achieve, and let algorithms find a way to meet these properties. The Step By Step Modeling Of PuzzlOr Electrifying Problem gives a detailed example of how to use optimization to solve a business problem. In order to make the use of optimization easier for data scientists, we have developed a python package for creating and solving optimization models. This package is available on PyPI as docplex. It can therefore be installed in your favorite Python environment with:
The documentation is available at: http There are few noticeable novelties compared to the beta version I used, including:
I could spend time writing about how good this modeling api is, but the best way to convince yourselves is to use it! By default the package uses either a local installation of CPLEX, or a subscription to DOcloud. The former is free for academics, while the latter has a free 30 days trial.
