Jump to: navigation, search

9.0.007.03

Journey Optimization Platform Release Notes

Important
Journey Optimization Platform was renamed to AI Core Services in release 9.0.009.01.


9.0
AI Core Services is part of 9.0 starting in 9.0.006.05.
Release Date Release Type Restrictions AIX Linux Solaris Windows
01/18/18 Update X

Helpful Links


What's New

This release includes the following new features and enhancements:

  • The Predictive Routing API has been extended to enable you to delete Agent Profile and Customer Profile schemas. The DELETE request should be made to the <base_url>/api/v2.0/schemas endpoint. The request body must include the API access token and a value for the schema_type parameter (agents or customers).
  • How Predictive Routing handles dataset field visibility has been updated to improve performance on the Datasets tab. Now, when a dataset is initialized, fifteen low-cardinality fields are made visible by default and displayed on the Datasets tab. You can add or hide fields by changing the visibility settings on the Settings: Datasets window. Only visible feature appear as facets and columns on the Dataset Details tab. However, regardless of their visibility status, all dataset fields are available for analysis.
     
    This functionality requires you to run the upgrade script provided in the Upgrade Notes section of this Release Note (below).


Resolved Issues

This release contains the following resolved issues:


The Agent Variance analysis report now uses Standard Error of the Mean, or SEM estimation, which measures how far the sample mean of the data is likely to be from the true population mean. The SEM is always smaller than the Standard Deviation. Previously, When the Agent Variance report was set to group by the same parameter as the Agent ID, the standard deviation sometimes could exceed the boundaries of the target metric. (PRR-1738)


The Lift Estimation analysis now correctly runs only on the test section of a dataset. Previously, it also incorporated data used to train the model it was evaluating, resulting in a flawed analysis. The train and test sections of the historical data are now split on the basis of time, rather than randomly, enabling Predictive Routing to identify the correct data to create the Lift Estimation analysis. (PRR-1694)


The Agent Variance analysis no longer fails when supplied with string data in place of numeric data.  (PRR-1693)


The Lift Estimation analysis report requires that there is an overlap between agent profile data and the predictor's dataset. Those agents are being scored to estimate if there is a lift. This fix addresses the issue when there is no overlap between Agent Profile and predictor's data by trying to score the agents from predictor itself. (PRR-1691)


Predictive Routing now runs correctly on Internet Explorer v11. Previously, it experienced the following issues:

  • All plots generated by analysis reports were distorted.
  • The Help widget did not work.

(PRR-1683)


Numeric data is now stored as floats rather than integers. Previously, all float values were saved as integers, introducing inaccuracies into the analysis reports. This functionality requires you to run the upgrade script provided in the Upgrade Notes section of this Release Note (below). (PRR-1675)


Validation has been added to the Number of Agents field in the Agent Variance analysis report to prevent incorrect values from being passed to the report. In addition, the default value of 50 has been added to the same field. (PRR-1672)


The Agent Variance report now generates a correct analysis when the target metric is of the boolean type. (PRR-1668)


You can now select None as the Group By parameter in the Lift Estimation analysis report, which means that Predictive Routing selects random interactions to calculate the estimated lift for agents. (PRR-1661)


Inactive, locked models are now correctly available for analysis in the Lift Estimation report. (PRR-1614)


Validation have been added to the Number of Samples and Number of Simulations fields in the Lift Estimation analysis report to prevent wrong values being entered. Also, these fields are both supplied with the default value of 100. (PRR-1612)



Upgrade Notes

Use the following scripts to upgrade JOP to release 9.0.007.03.

The following script activates the new dataset fields visibility feature. To install the upgrade, run the following script in the Tango container:

docker exec -it tango /bin/bash
cd solariat_bottle/src/solariat_bottle/jop/common/scripts/versioning
python upgrade_35a_prr_dataset_visible_fields.py --mode=prod

The following script activates the correction to how JOP handles float values. To install the upgrade, run the following script in the Tango container:

docker exec -it tango /bin/bash
cd solariat_bottle/src/solariat_bottle/jop/common/scripts/versioning
python upgrade_34a_prr_integer_vectorizer.py --mode=prod
Important
This script was formerly named upgrade_34a_prr_integer_vectorizer.py. The change in the number from 34a to 34b did not affect the contents of the script. Aside from the name, both versions are identical.
This page was last modified on March 29, 2018, at 07:54.

Feedback

Comment on this article:

blog comments powered by Disqus