Building Decision Optimization models

IBM® Decision Optimization gives you access to IBM's industry-leading solution engines for mathematical programming and constraint programming. You can build Decision Optimization models either with notebooks or by using the powerful Decision Optimization experiment UI. Here you can import, create, or edit models. To create or edit your models, you can use Python, OPL, or the natural language expressions that are provided by the intelligent Modeling Assistant. You can also solve multiple scenarios and save models to deploy with the Watson Machine Learning service all from within the experiment UI.

Service This service is not available by default. An administrator must install the service. To determine whether the service is installed, open the Services catalog. If the service is installed and ready to use, the tile in the catalog shows Ready to use.

Data format
Tabular: .csv, .xls, .json files. See Preparing input data in a Decision Optimization experiment

Data from Connected data assets

For deployment, see Model input and output data file formats (Decision Optimization)

Data size
Any

See also Deploying Decision Optimization models with Watson Machine Learning.

Accessing Decision Optimization

To create a Decision Optimization experiment, follow these steps.

  1. Open your project or create an empty project.
  2. Select the Assets tab.
  3. Select New asset > Solve optimization problems in the Work with models section.
  4. Click New deployment space, enter a name, and click Create (or select an existing space).
  5. Enter a Name for your Decision Optimization experiment and click Create.

The Decision Optimization experiment UI opens where you can create and edit models that are formulated with the Modeling Assistant, or in Python DOcplex, or in OPL.

Alternatively, to open and run Decision Optimization notebooks (without the Decision Optimization experiment UI), follow these steps.

  1. Select the Assets tab.
  2. Select New asset > Work with data and models in Python or R in the Work with models section.

For a step-by-step guide to build, solve and deploy a Decision Optimization model, by using the user interface, see the Quick start tutorial with video.

What is Decision Optimization?

People frequently use the term optimization to mean making something better. Although optimization often makes things better, it means a lot more: optimization means finding the most appropriate solution to a precisely defined situation. It is a sophisticated analytics technology, also called Prescriptive Analytics, which can explore a huge range of possible scenarios and suggest the best way to respond to a present or future situation.

Decision optimization

  1. The situation is generally a business problem, such as planning, scheduling, pricing, inventory, or resource management.
  2. Whatever the problem is, resolving it starts with the optimization model, which is the mathematical formulation of the problem that can be interpreted and solved by an optimization engine. The optimization model specifies the relationships among the objectives, constraints, limitations, and choices that are involved in the decisions. But it is the input data that makes these relationships concrete. An optimization model for production planning, for example, can have the same form whether you are producing three products or a thousand. The optimization model plus the input data creates an instance of an optimization problem.
  3. Optimization engines (or solvers) apply mathematical algorithms to find a solution, a set of decisions that achieves the best values for the objectives and respects the constraints and limitations imposed. The optimization engine implements specialized algorithms that are developed and tuned to efficiently solve a large variety of different problems. Decision Optimization uses the IBM CPLEX and CP Optimizer engines that have proved powerful in solving real-world applications.
  4. The solution that emerges from the solver details the recommended values for all decisions that are represented in the model. Equally important are the metric values that represent the targets. These values measure the quality of the solution in terms of the business goals.
  5. All of this can be made available to business users with a complementary business application. Usually, the objective and solution values are summarized in tabular or graphical views that provide understanding and insight.

Learn more