Global Data Science Forum

 View Only

Decision Optimization Now Available in the Watson Machine Learning Service (IBM Cloud Blog)

By Juliet Sigmann posted Mon June 24, 2019 02:24 PM


Check out this blog posted originally on IBM Cloud Blog here:

Decision Optimization Now Available in the Watson Machine Learning Service

By: Virginie Grandhaye

We are excited to announce that CPLEX engines are available in the IBM Watson Machine Learning (WML) service as of June 21, 2019.

The Gartner forecast says that the prescriptive analytics software market will reach $1.88 billion by 2022, with a 20.6% CAGR from 2017.

IBM Decision Optimization provides prescriptive analytics capabilities by leveraging powerful CPLEX engines to transform your predictions into actionable plans. It answers the question “What should I do?” and is a good complement to what machine learning models can predict (“What will happen?”).

Using the power of mathematical programming or constraint programming techniques, you can model your business problem and use it for decision support applications. CPLEX engines can explore many possibilities and provide the optimal solution to a problem that would otherwise take days or weeks for a human brain to solve (without being sure to get the optimal one).

In other words, you use mathematical algorithms to sift through all possible solutions and recommend the best one that optimizes a business goal, taking into consideration all decision variables, constraints, and trade-offs.

It applies in many industries and can propose solutions for problems like portfolio optimization, pricing optimization, staff scheduling, and resource allocation problems.

You can now deploy and solve your Decision Optimization models inside the Watson Machine Learning (WML) service

For each deployment in the Watson Machine Learning service, you can specify the run capacity you want to allow to this deployment (# VPC, # RAM). You’ll also have the ability to specify how many pods you want to allocate to this deployment. This capability will then allow you to queue your jobs and benefit from load balancing on your dedicated pods. You can connect your model to external data sources (e.g., IBM Cloud Object Storage, IBM Db2 on Cloud) or just provide your data inline.

Application developers will benefit from WML V4 API

Developers will be able to get an endpoint for predictions—provided by the machine learning model—and an endpoint within the same service, for prescriptive results, provided by the Decision Optimization model.

This new capability in the Watson Machine Learning service will give you the ability to just consume CPLEX engines without installing anything on your laptop. Watson Machine Learning delivers an easy-to-use, self-service decision environment for your organization. The solution helps all users harness the power of optimization-based decision support without the install, deployment, and maintenance requirements of traditional, on-premise infrastructures.


The notebook "Deploying a Decision Optimization Model with Watson Machine Learning" provides an example of how to deploy your model.

Find out more

The service will be updated in all available regions. To learn more, visit the IBM Watson Machine Learning page.