This release is a minor update release in the 4.x series
Updates
Adds support for DataRobot API version 2.20, while maintaining backwards compatibility with version 2.19
v4.1.4 (December 10, 2019)
This is a minor update release in the 4.x series
Updates
Removed proxy settings from both the DataRobot Predict Tool as well as the DataRobot Automodel Tool.
Predictions with dedicated prediction servers are being deprecated. Removing the ability to create new dedicated prediction server workflows, though previously created workflows will continue to be supported.
v4.1.3 (August 7, 2019)
This release is a minor bug-fix release in the 4.x series:
Fixes
Fixed a bug causing communication over the wrong port when "Disable SSL" is selected.
v4.1.2 (June 5, 2019)
This release is a minor bug-fix release in the 4.x series
Fixes
Fixed bug when incorrect DataRobot API Token is entered for validation.
Enhancement
Added information to help document about correct encoding for non-Latin characters in datasets
v4.1.1 (May 10, 2019)
This release is a minor bug-fix release in the 4.x series
Fixes
On prem prediction requests using deployments will now use the correct http protocol based on the internal url entered when logging in.
v4.1.0 (May 1, 2019)
This release is a minor enhancement in the 4.x series.
Enhancement
Binary classification predictions made using deployments in the DataRobot Predict Tool now return the same positive and negative
classes used during modeling, rather than returning 'True' or 'False'. Binary classification predictions made using models will return 'True' or 'False' as before.
v4.0.2 (April 25, 2019)
This release is a minor bug-fix release in the 4.x series
Fixes
Predictions against models built using string binary classes (eg. 'yes' or 'no') will now correctly return True or False instead of only False.
v4.0.1 (April 17, 2019)
This release is a minor enhancement in the 4.x series.
Enhancement
Added Relax known in advance features selector for time series deployments in the DataRobot Predict Tool. Selecting this will allow for missing known in advance features in prediction rows at the cost of lower prediction accuracy.
v4.0.0 (April 12, 2019)
This is a major release in the 4.x series. From this version onward,
we officially support only 2018.X and 2019.X versions of Alteryx.
Why Upgrade
DataRobot Predict Tool now supports deployment based predictions.
Deployments allow model health monitoring and drift tracking, and
bring support for time series models. Additionally, interface updates
make both tools more pleasant to use.
Enhancements
Add deployment based predictions.
Add support for Time series predictions.
Add output features support to be included in predictions results.
Add option to upload data only, but skip selecting target and launching autopilot.
Improved interface of both tools.
v3.2.1 (July 11, 2018)
This release is a minor enhancement in the 3.x series.
Enhancements
Improved support for large datasets in the Automodel tool (Alteryx 2018.1+ only).
Save our support log in %AppData% if the user doesn't have permission
to write to %ProgramData%.
v3.2.0 (March 30, 2018)
This release is a minor enhancement in the 3.x series.
Changes
Changed the location of where the support log gets written to make it
consistent between Designer and Server installs. See the documentation for
details.
Enhancements
We now officially support running these tools on Alteryx Server 11.7 and later.
We now rotate our support log file so the setting should be safe to keep enabled
for extended periods of time without the risk of filling up ones disk.
v3.1.0 (March 21, 2018)
This release is a minor enhancement in the 3.x series.
Enhancements
Added support for Alteryx 2018.1. Important: when upgrading Alteryx
(Server or Designer) the DataRobot tools will need to be re-installed after
upgrade to function properly.
We now provide more helpful error messages when we timeout while computing
predictions. We will direct users to adjust their Batch Size setting.
Model details display in Predict tool are now more concise.
Fixes
In some situations, DataRobot will return less than the number of reason codes
that the user requested. The old tools would raise an AssertionError but we
now correctly deal with this case and fill in missing values with null.
v3.0.4 (March 20, 2018)
This release is a minor enhancement in the 3.0.x series.
Enhancements
A new Advanced Option was added to the Predict Tool, Batch Size, that allows users
to have more control over the size of the batches that the tool chunks its prediction
data into.
v3.0.3 (March 9, 2018)
This is a minor bug-fix release in the 3.0.x series.
Enhancements
A new Advanced Option was added, Create Support Log, that when enabled,
will create a special log file that can be sent to a DataRobot Support
Technician to aid in troubleshooting customer issues. Read more about it in
the tool's help documentation.
Fixes
Fix handling of null values in Reason Code output: the field "Reason N Value"
used to contain the string "None" when the given feature was null for that
row. The field will now be a true null value.
Fix issue with getting predictions from a Binary project when using an older
version of DataRobot.
Fix issue with getting predictions from a Binary project when the positive and
negative classes are custom.
More friendly error message when a user hasn't selected a Project and they try
to run the workflow.
v3.0.2 (February 20, 2018)
Minor changes to the documentation included in our tools. Also, better validation
of prediction results as returned by DataRobot.
v3.0.1 (February 13, 2018)
This is a minor packaging change in the 3.x series. If upgrading from 2.x
please see the v3.0.0 entry for more details.
v3.0.0 (February 5, 2018)
This is a major release in the 3.x series. This release supports Alteryx
versions 11.7+ and newer. For support of older versions of Alteryx, please use a
connector from the 2.x line of releases. This connector supports DataRobot
Enterprise versions 2.9+ and the Cloud platform.
Why Upgrade
With this v3.0.0 release, predictions and reason codes are are much faster to generate,
and support much larger loads, especially when a dedicated prediction server is used.
Use for Enterprise installs has also become easier, with the ability to work with
firewalls and a new beta rollout of support for proxy server connections. Additionally,
new models are now possible -- one can now train and predict multiclass models, such as
predicting which deal (among several possible deals) is best for a customer. Lastly, a
lot of the UI has been streamlined to make input and output easier to use and understand.
Changes
The Automodel tool now outputs the Project ID and ID of the best performing
model that was trained.
Output from the Predict tool will no longer include the original
input. It is recommended that you use the built-in Alteryx Join Tool to
join the prediction results with the desired fields from the original data.
Output field names of the Predict tool are now dynamically named based on
the project type.
Authenticating via a password is no longer supported. The only supported
authentication mechanism is with the DataRobot API Token which can be found
in the Profile page of the DataRobot web UI.
This suite of tools no longer requires Alteryx Predictive Tools to be
installed.
The icons have been changed to visually differentiate the legacy tools from
this suite.
Enhancements
Requesting predictions from Dedicated Prediction Servers is now supported.
Getting predictions from Multiclass models is now fully supported.
Creating Multiclass projects from the Automodel tool is now supported as
long as your target variable is unambiguous.
Output field names have been improved to be more informative and include
short descriptions in the metadata view.
Large datasets are now better supported.
Reason Codes output fields have been simplified.
Users can now choose to ignore SSL certificate errors. This option can be
useful for Enterprise installs and for customers behind certain forms of
firewalls.
Users can configure HTTP/S proxy settings (BETA).
Inline documentation is included in the tool.
This is the first release to leverage the Alteryx Python SDK for improved
performance and reliability.
Fixes
We support creating projects up to the DataRobot limit and requesting
predictions up to the limits of your machine.
Error messages are more informative.
Empty Reason Codes rows (NULL) now output correctly.
v2.0.2 (January 15, 2018)
This is a minor bug-fix release in the 2.0.x series
Fixes
Fix warning in Automodel tool about the App/Macro being out of date.
v2.0.1 (January 15, 2018)
This is a major release in the 2.x series. This change continues to support
Alteryx versions back to 10.5; however, new features will only be supported
in 11.5+. Also, this connector now supports DataRobot Enterprise versions
3.0+ and the Cloud.
Enhancements
Add option to return Reason Codes in the Predict tool. Warning: enabling this
option will change the format of the output columns.
Makes all the drop down fields searchable.
Other minor UI tweaks.
Use new Alteryx HTML SDK v2 if it is available.
Update the embedded DataRobot R client to 2.6.0. This version is compatible
with DataRobot Enterprise 3.0+ and the Cloud.
Rudimentary support for MultiClass projects was added to the Predict tool
only. The bestPredictions column will contain the predicted class but no
probability information will be returned.
Fixes
Changing credentials to a different DataRobot server no longer breaks the
Predict tool UI.
Failed runs will properly cleanup the uploaded data on the DataRobot server.
Setting max workers beyond the account limit is now a warning rather than an
error.
v1.1.0 (November 7, 2017)
This is a minor release in the 1.1.x series. The connector is now officially
compatible with DataRobot Cloud and Enterprise 2.9+.
Enhancements
Adds Connector version string to configuration UI. The string is a clickable
link which will will display the full list of changes in each release.
Record the Connector version that was used when the workflow was last saved.
Embeds DataRobot R client (2.4.0) into this package. Upon installation of
this tool, the R-library folder will be copied into the user's library path.
DataRobot R client 2.4.0 is compatible with DataRobot Cloud and Enterprise
2.9+. MultiClass projects are not supported in either the Automodel or
Predict tools.
Predict and Automodel tools now connect to the target DataRobot server with a
custom User-Agent header to aid in troubleshooting.
v1.0.7 (October 20, 2017)
This is a minor bug-fix release in the 1.0.x series
Enhancements
Output more informational messages during modeling and prediction steps
Better error messages for network connectivity issues
Fixes
Fix issue where Automodel and Predict tool were silently renaming column
headers with non-standard characters (i.e. - : % & etc) with an
underscore (_)
v1.0.6 (October 18, 2017)
Fix packaging issue from prior release
v1.0.5 (October 9, 2017)
This is a minor bug-fix release in the 1.0.x series
Fixes
The DataRobot Server error message is passed correctly to Alteryx from R
The project and model drop-downs will correctly display the project and model
names and not just their IDs
v1.0.3 (September 13, 2017)
This release fixes a UI issues with Alteryx Designer 11.3+
Fixes
Fix the credentials configuration page for the DataRobot Automodel tool
Fix the credentials configuration page for the DataRobot Predict tool
v1.0.2 (June 12, 2017)
This is a minor bug-fix release in the 1.0.x series
Fixes
The DataRobot Predict connector responds to edited Instance URL