Data Workbench 6.6 Release Notes

Data Workbench 6.6 release notes include new features, upgrade requirements, fixed bugs, and known issues.

New Features

Data Workbench 6.6 includes these new features:

Smart Feature Selection

The Smart Feature provides automated feature selection when using the Propensity Scoring visualization. You no longer need to select the independent variables from a long list of metrics and dimensions.

Select Options > Feature to enable the feature selection menu. If the Feature Reduction option is not selected, you can make the system explore additional model inputs from the entire set of metrics while retaining user-added metrics. Checking Feature Reduction will make the system select the most relevant metrics from the user metrics.


While using Smart Feature Selection for an empty model, Feature Selection Settings are redundant.

Selecting More Features or Less Features from the Level Setting menu makes the selection algorithm add feature sets to the model. In some cases both setting types may return the same model.
Segment Export Wizard The [Segment Export Wizard] provides a step-by-step process to configure and export segments.

New export executable for Target exports

To export Adobe Target data, a new TargetBulkUpload.exe has been developed to replace the ExportIntegration.exe for Target exports. If a user wants to move legacy exports to TargetBulkUpload.exe, then exports have to be changed (change mbox3rdPartyId to thirdPartyId).

L4 Logging

An option for L4 Logging has been added under Server > Admin > Export menu ,as well as a new configuration file (loggingDetails.cfg). This allows you to turn GE and L4 logging on and off. By default, L4 Logging will be set to true and GE Logging will be set to false in the configuration file, but you can change these settings as needed. This is only for exports using the exportintegration.exe.

New Presentation Layer options

You can now set transparency and border thickness for Circle and Rectangle shapes within the Presentation Layer. Right-click on the object and select No Fill to make it transparent. Four border thickness options are present under Outline—Thin, Medium, Thick, Thickest.

New options for Correlation and Association Chord diagrams

New grouping options have been added in the Correlation Chord and Association Chord diagrams. The right-click menu now includes Add Group, Remove Group, and Add to Group.

Grouping will display once you populate your chord. All elements part of the same grouping will be based on the group’s color. You can set the group’s color like you would set an element’s colors. If a grouping is deleted, then the elements are moved to Ungrouped.

Set transparency and border thickness for annotations

You can now set the transparency and border size of circles and rectangles to highlight and annotate visualizations in a workspace. New options for the circle and rectangle include a No Fill option for transparency and border thickness settings of Thin, Medium, or Thickest.

XML Decoder enhancements

You can now use the #value tag in XML paths to pull the value of an XML element. For example, you can now read the attribute value of the <Hit><Page name="Home Page" index="20">home.html </Page></Hit> tag using a Hit.Page.#value expression. See XML Decoder Groups for additional information.

New features in Propensity Score

Within the toolbar button you can now choose to either Save a filter or automatically launch Propensity Scoring with a new feature selection capability. The Propensity Scoring opens the builder with the workspace selections applied for the dependent variable, enables the selected features, and lets you hit Go to start.

It is now possible to save a Propensity Score Model as a JavaScript file. After the model is built click on Save > Save JavaScript.

Additional Visit level predictive builders

Added new predictive builders with Visit as the root countable dimension for Propensity Scoring, Clustering, and Decision Trees.

Additional Latency visualizations

Added a set of Latency visualizations. The former latency visualizations looked at an equal period around the event (-7 to +7 days) to create an assignment. This works well for only one event per clip, but doesn’t give adequate results when there are multiple events, such as multiple conversions.

We have reorganized and added additional latency visualizations to the menu that are configured to evaluate latency before an event, after an event, and before and after a single event.

Updates to report.cfg

You can now view dates in calendar format in the report.cfg configuration tree.

Chord Visualization updates

Implemented the ability to scale in the Chord Visualization and Pie Chart visualizations. You can scale by holding down <Ctrl> key and using the mouse wheel, or by right-clicking on the window border, selecting a value to scale, and entering a value.

Access to Detailed Status

Sensor 5.0 supports:

  • 2048 bits certificates signed using SHA256.
  • Support for Apache Server 2.4 on Linux, Sun Solaris, or FreeBSD
  • Support for Apache Server 2.4 on Windows Server 2000 or later

Access to Detailed Status is now restricted to Power Users and Administrators by default.

System Updates

These features have been renamed, deleted, or the installation files or folders were restructured in this release:

  • Default extension for a reportserver.cfg file is now .xlsx.

  • Additions to Insight.cfg (the configuration tree).

    • printf format: You can now set the printf format from the Insight.cfg tree.

    • V3D: You can now set a V3D setting using three associated edit boxes.

    • Date and Time: You can now set the date and time from a dropdown menu.

  • The Page Overlay feature has been removed from the options menu for a URL table element. If you have an older workspace with a page overlay visualization and click on the Refresh Page menu option, a dialog will appear stating: This feature has been deprecated due to security concerns.

Fixed Bugs

The following are prominent fixes made in Data Workbench 6.6 (since the release of Data Workbench 6.52).

  • Fix to Best Fit Attribution (algorithmic attribution): Coefficients were are all zeros when SGD was used to converge. This has been fixed.

Known Issues

  • The targetbulkupload.exe requires command-line arguments in English. The output file name should be in English for Adobe Target Export, Profiles and Audiences Export, and Customer Record Service Export.

On this page