By: Jakub Hava
There are several new features in the latest Sparkling Water. The major new addition is that we now publish Sparkling Water documentation as a website which is available here. This link is for Spark 2.2.
We have also documented and fixed a few issues with LDAP on Sparkling Water. Exact steps are provided in the documentation.
Bundled H2O was upgraded to 184.108.40.206 which brings the ordinal regression for GLM as the major change.
The last major change included in this release is the availability of the H2O AutoML and H2O Grid Search transformation for the Spark pipelines. They are now being exposed as regular Spark Estimator and can be used within your Spark pipelines. An example PySparkling AutoML script can be found here.
The full changelog can be viewed here.
Sparkling Water is already integrated with Spark 2.3 on master and the next release will be also for this latest Spark version.