AI Core Services Release Notes
|Release Date||Release Type||Restrictions||AIX||Linux||Solaris||Windows|
|08/23/18||General||Under Shipping Control||X|
This is a general release for this component. For availability of this release, contact your Genesys representative. This release includes the following new features and enhancements:
- The Quality column in the Models list table on the Model configuration window now includes a new metric, Local models. The metric displays the number of local models generated for agents in the dataset on which the predictor is built. Local models are built only for models that have the HYBRID or DISJOINT type. For GLOBAL models this metric is always 0. If a model is a new (that is, inactive and untrained) the metric value is -1 until the model is trained. Once trained, the metric shows the actual number of local models. (PRR-3120)
- AI Core Services (AICS) now supports Oracle Enterprise Linux 7.3. (PRR-3013)
- AICS has improved handling of UTF-8 characters. Data ingestion, model training, and analysis reports are all correctly processed for data containing non-ASCII UTF-8 characters.
- ImportantNames of predictors and features must still contain ASCII characters only.
- The GPR API now enables you to run Feature Analysis reports. The API returns a JSON response containing a list of features ordered by weight--that is, by the strength of the impact that feature has on the value of the target metric. The resulting report is also automatically available for view from the GPR web application. Your API request should include at least the dataset ID and the target metric (KPI) column name. You can also specify columns to exclude from the analysis, which enables you to have GPR omit columns that are not applicable for the analysis, such as Employee ID, Name, and so on. For more information, see the Predictive Routing API Reference. (This file requires a password to open it. Contact your Genesys representative if you need access.) (PRR-2560, PRR-2565)
- The GPR API now enables you to run Lift Estimation reports. The API returns a JSON response containing the Lift Estimation results. The resulting report is also automatically available for view from the GPR web application. The Lift Estimation results are presented as an actual metric versus a predicted (estimated) metric. A sample response is shown below:
- For more information, see the Predictive Routing API Reference. (This file requires a password to open it. Contact your Genesys representative if you need access.) (PRR-2557)
- The Lift Estimation report now uses the scoring expression configured for the predictor (if any) to decide whether the target metric should be minimized or maximized, enabling GPR to match an interaction with the agent having the optimal score. If you have not configured a scoring expression for your predictor, GPR assumes it is a maximization metric. See Understanding Score Expressions for a detailed discussion of this functionality and how the Lift Estimation graph presents metrics that should be minimized versus those that should be maximized. (PRR-2010)
This release contains the following resolved issues:
The Feature Analysis report now supports a target metric with float values in your dataset. Previously, if the target metric was of the float type, its value was automatically converted to the integer type, which could lead to analysis errors. (PRR-3268)
GPR now correctly handles UTF-8 dataset, agent profile, and customer profile files that contain a Byte Order Mark (BOM), which is automatically inserted by most Microsoft applications. Previously, such files were incorrectly processed and could not be used for analysis or building a predictor. (PRR-3211)
The time required to append data to a dataset has been improved by changing how cardinalities are computed. Cardinalities are now computed only on the appended data and the resulting cardinality values are added to those already stored in the database. If you need to have GPR recompute cardinalities for the entire dataset, use a PUT request to the GPR API .../datasets/<dataset_id>/compute_cardinalities endpoint. Genesys recommends that you recompute cardinalities after deleting or purging data from the dataset. For more information on the API, see the Predictive Routing API Reference. (This file requires a password to open it. Contact your Genesys representative if you need access.) (PRR-3170)
The endpoint enabling you to generate predictor data using the Predictive Routing API now accepts a time range parameter. (PRR-2962)
No special procedure is required to upgrade to release 9.0.012.01.