Jump to: navigation, search

AI Core Services Release Notes

AI Core Services is part of 9.x starting in
Release Date Release Type Restrictions AIX Linux Solaris Windows
01/25/19 General Under Shipping Control X

Helpful Links

What's New

This release includes only resolved issues.

Resolved Issues

This release contains the following resolved issues:

The table functionality on the Details tabs of the Datasets, Agents, Customers, and Predictors windows has been improved. Previously, rows not initially visible within the window appeared as you scrolled down. Any newly-added data appeared at the end of the table without the page refreshing, which sometimes led to slow page rendering and memory leaks. The scroll-based page display has now been replaced with page-by-page navigation, where each page displays only a limited number of records. (PRR-4170)

The line graph on the Predictor Trend tab has been replaced with a histogram chart. (PRR-4167)

Local Models are now correctly generated for all agents who have enough data (at least one record if you are using a numeric target metric or one record for each class of a Boolean metric) and whose records fall within the training portion of the dataset after it is split into the training and testing sections. Previously, for Datasets larger than one million records, only agent records occurring in the first million rows were used to build Local Models. (PRR-4130)

Requests to the Feature Analysis report API endpoint now correctly handle valid attributes. Previously, in some cases, a valid attribute was flagged as invalid. (PRR-4041)

If a synchronization error occurs on a Dataset containing non-ASCII column names or column names with spaces, the error message now displays the actual column name. Previously, the error message displayed the hash string used to represent the column name. (PRR-4031)

You can now add a simple predictor to a composite predictor and immediately add the simple predictor to the composite predictor expression. Previously, you had to re-save the composite predictor before adding the new simple predictor to the expression used for the composite predictor. (PRR-4010)

AICS now correctly connects to the new master MongoDB database when a master re-election occurs. Previously, AICS continued to try to connect to the former master database instance. (PRR-3952)

In the GPR web application, Agent and Customer features with very long names are now displayed correctly. Previously, these names might extend beyond the margins of the text boxes allocated for them or might be truncated where the feature name extends past the allotted text area. (PRR-3943)

You can now create a Model via the GPR API that does not include Agent features. Agent features are not required to create a Model, but previously, the API did not allow you to create a Model without them. (PRR-3829)

Upgrade Notes

After upgrading to release, you might notice that old jobs (such as data uploads or analysis report creation) are in the Pending state, preventing new jobs from running. To clear out the old jobs, use the following workaround:


  1. Connect to the MongoDB database using the following command:
    $ docker exec -it mongo mongo --ssl --sslAllowInvalidCertificates localhost:27017/solariat_bottle
  2. Drop the jobs collection using the following command:
This page was last edited on March 1, 2019, at 14:05.
blog comments powered by Disqus