Table of contents

What's new in Decision Optimization for Watson Studio

See what's new in Decision Optimization for Watson Studio.

What's new in Decision Optimization for Watson Studio Version 1.2.3 (February 2019)

Python version

Decision Optimization for Watson Studio V1.2.3 currently supports Python V2.7. This is the last release to support this version of Python. The next release of Decision Optimization for Watson Studio will no longer support Python V2.7, but will support Python V3.x.

What's new in Decision Optimization for Watson Studio Version 1.2.2 (October 2018)

Product name change

IBM Decision Optimization for Data Science Experience is now IBM Decision Optimization for Watson Studio.

IBM Deployment Manager is now IBM Watson Machine Learning.

What's new in Decision Optimization for Data Science Experience Version 1.1 (August 2018)

Improved editing in the Model Builder

  • You can preview tables in the Select Data view before importing them into your scenario.
  • Changes made to the Project Data sets will be displayed in the Select Data view, but these will not have impact on the scenario tables in the Prepare Input Data view, unless you choose to re-import them into the scenario.
  • Changes made in the Prepare Input Data view are saved in the scenario, but have no impact on the Project Data sets displayed in the Select Data view.
  • You can now re-size columns in the Prepare Input Data view and in tables in the Dashboard.
  • You can also add and remove rows in the full screen mode of the Prepare Input Data view.
  • You can now create and edit Python models in the Model Builder with improved syntax support and some highlighting.

OPL models

OPL models can now be imported and run in the Model Builder. When formulating your model, select Import a Model and then choose From File to select your OPL model.

Scenario management

  • You can now export scenarios as a zip file, see Exporting scenarios. This can be useful if you want to use a scenario in another project.
  • You can now import scenarios as a zip file, see Importing scenarios. New scenarios can be imported by selecting Create scenario from file from the Scenario panel. This can be useful for debugging if you have used run configuration parameters (see Custom Parameters) to generate a debug zip file. The debug zip file will provide you with a scenario containing data, model, solution and the run configuration parameters.
  • You can now switch scenarios while running a model and run models simultaneously (depending on your subscription: see CPU core and memory used in scenarios).

New version of Modeling Assistant

The beta version ended 10 August 2018.

New decision domains in Modeling Assistant

  • Selection & Allocation problems can now be formulated in the Modeling Assistant.
  • Supply & Demand Planning problems can now be formulated in the Modeling Assistant.
  • To help you choose the most appropriate decision domain, a new wizard helps you to decide, see Selecting a Decision domain in the Modeling Assistant.
  • Other decision domains can be easily imported for industry problems supported by IBM Business Partners. For more information contact your IBM Sales representative.

Easier model formulation in the Modeling Assistant

  • Constraints or objectives that you need to be completed are more clearly highlighted in red.
  • The suggestion filter has been redesigned.
  • You can now duplicate statements, or move statements up and down by clicking the 3 vertical dots located next to each statement.
  • Show/hide arrows appear at the beginning of statements which have associated rules.
  • Scale factors can be added to the objective function, either by entering it as a literal number, or as a path to the data. See also Solution view improved.
  • Precedence constraints can now be configured with more flexibility by specifying the type of each precedence as an entry in the input data.
  • You can now search objectives and constraints and the results are highlighted in your model.
  • Constraints containing periodic time intervals defined by days and times (for example every Friday 7pm to Monday 9am) are now possible.

New run configuration parameters

  • You can now specify Run configuration parameters. See Run Configuration Parameters and Decision Optimization environment for details on these parameters.
  • You can also use these new parameters to change your Decision Optimization environment for each scenario, to enable you to run models simultaneously.
  • You can add some conditions on the solve, for example, if the model has exceeded memory or solve time limits, a log file is dumped.
  • Your run configuration parameters are saved when you save your model for deployment. These parameters are subsequently used in the deployment service.

New graphics displayed during Run

Feasible solutions obtained during the run are now displayed graphically until the optimal solution is found.

Solution view improved

Weights added to objectives are now shown as scalings in the Solution view for models built using the Modeling Assistant. See Multi-objective weights.

New dashboard features

  • New types of chart widgets are now provided in the dashboard.
  • Table columns can be re-sized.

Easy table filtering

You can filter table values using the search field. See Table search and filtering.

dsx-samples

The dsx-sample Notebook DecisionOptimizationExecuteDeployedModel.ipynb has been updated to show you how to activate the dump of a zip file containing the input and output data, model and run configuration parameters. This can be useful for debugging purposes. The sample also illustrates how to retrieve the zip. After debugging you can upload this zip into a scenario. See Submitting a job to a deployed model.

Two new samples PorfolioAllocation and SupplyDemandPlanning are now available. These illustrate the new Selection and Allocation, and Supply and Demand Planning decision domains that are now supported by the Modeling Assistant.

The MarketingCampaignAssignment sample, which is a Resource Assignment example, now also contains an extra scenario (Scenario 4 - Selection) which shows you how to model this problem using the Selection and Allocation domain. (This is possible because the possible candidates to select from are listed in the same data table).

For a list of all samples provided see Decision Optimization Models.

Upgrading to latest version of DSX Local

After upgrading from DSX Local V1.2.0.3 to V1.2.1, if you access an existing project, you are prompted to commit changes. This is necessary to ensure that your project is migrated correctly in the latest version of DSX Local. (In this migration, files are moved from the decision-optimization folder to a domodel folder.) You must simply click commit. This is also the case if you import a project that you had exported previously in an earlier version.

Models that you have deployed in V1.2.0.3 will continue to run, but in the new version V1.2.1 you will no longer see your deployed models in the Asset view. This is normal as DSX Local only lists models that are deployed in the current version, but your model is still running. If you have no changes to make to your deployed model, no action is required. When you eventually change your model and redeploy the updated project, it will appear again in the Asset view.