Technology Blogs by SAP
Learn how to extend and personalize SAP applications. Follow the SAP technology blog for insights into SAP BTP, ABAP, SAP Analytics Cloud, SAP HANA, and more.
Showing results for 
Search instead for 
Did you mean: 
0 Kudos
Release 2.2 of SAP Predictive service is now available.

What's new in Release 2.2

Services improvements

Improvements in the performance, of the security and fix devops bugs.

APL is now supported.

Previous releases

Release 2.1

Services improvements

All services:

Key Influencers: The service now allows end-users to get the variables that are correlated to each other, with their coefficient of correlation (see Response Body Parameters).


  • An important note has been added to the Clustering APIs documentation to explain the link between the view export method and the modelSQLExportEnabled input parameter (see Clustering Request Body Parameters).

  • A correction has been made to the data source binding procedure. Only <default> can be used as data source name (see Bind the Data Source).

Release 2.0

Services improvements

The new collection of services Predictive Analytics Integrator Services is available to add model management tasks to your application. See Service Description.

This brings the possibility to train an Automated Analytics model on a registered dataset and to apply this model on a new dataset to obtain predictions on the target variable.

The steps to update from a previous 1.x release are the same. There are only 2 differences mentioned below, but a more complete description is given here.

  • Deploy the SAP Predictive service is identical

  • On your Database Systems, install SAP HANA Component SAP Predictive service Engine becomes mandatory each time there is a new version of APL and/or a new version of Predictive Analytics Integrator Services.

  • The technical database user must be granted role hana.pai::ExecutePAI to be able to use Predictive Analytics Integrator Services.

Release 1.14

Services improvements

Clustering service:The service now allows end-users to specify the distance used to measure the proximity of two data points. This is enabled through the distance input parameter. See Clustering APIs.

Forecast service:The service now allows end-users to:

  • Specify the maximum lag to consider when forecasts are computed. This is enabled through the maxLag input parameter

  • Get the MAPE indicator for each horizon. This is output in the mapePerHorizon parameter under modelPerformance

See Forecast APIs for more informations.

Release 1.13

Services improvements

Dataset service: The service now allows application users to specify the schema and the table/view of the dataset in SAP HANA separately in the request. This is enabled through the location input parameter. The hanaURL parameter has been deprecated. Refer to technical help for details.

Release 1.12

Services improvements

Scoring Equation: The service now allows application users to choose the type of output generated by the scoring equation (predicted value or probability). This is enabled through the predictionOutputType input parameter. See Scoring Equation APIs.

Release 1.11

Services improvements

Outliers and Scoring Equation: now allow application users to set the key of the target variable through the TargetKey input parameter.

All services: The variableDescription parameter has been deprecated from the request body of the APIs. From now on, application users must use the Dataset APIs to specify the variable descriptions or to modify them.


The documentation now specifies that referenceDate must follow the ISO 8601 format in the Forecasts API request.

Release 1.10

Clustering service

This is a new service which segment your dataset into homogeneous clusters and export the segmentation results into a SAP Hana table or view.

The scenario of this service is:

  1. Call clustering service to segment the dataset into homogeneous clusters

  2. Debrief detected clusters

  3. Assign a cluster to each element of the population.


Version of APL is now supported. Go to menu Persistence/Database Systems, select your database system and click on “Check for updates” button to get this new version of APL.

Services improvements

Outliers: The service now allows application users to enable autoselection of variables through the autoSelection input parameter.


Improvement and optimizations of existing services


Documentation is updated to reflect these enhancements and new features.

Release 1.9

New name

The name of the services follow the new naming convention of the cloud platform. So the new name is: SAP Predictive service.

Forecast service

The developer has now the choice between three modeling techniques to generate the forecasts. this implies the new input parameter “forecastMethod” when this service is posted:

  • “defaut” configuration used in Automated Time Series Analysis

  • “smoothing” technique which generates forecasts by smoothing a time series. The cycle length to use is mentioned is the parameter “smoothingCycleLength”. It is an integer whose default value is null.

  • “linear regression”

Dataset service

When a dataset is registered, it is now possible to flag a variable column as a component of a primary key.

SAP API Business Hub

SAP Predictive service can be explored and tested with 4 datasets from the SAP API Business Hub.


Improvement and optimizations of existing services.


A link has been added to the Overview to point to videos explaining how to deploy Predictive.

Release 1.8

Outlier service

Now this service detects about 2% or less outliers in the training dataset.

Dataset service

Modify the description of the variable of a dataset. Enable a user to correct the description of the variables of the dataset if what was determined by the guess is not correct. For example, forecast service requires the target column to be continuous. If it's not, continuous type can be forced to continuous.


Improvement and optimizations of existing services.

The Predictive service supports SAP HANA 1.0 SPS12.

Release 1.7

Forecast Service

When this service is called, it is now possible to request in the input parameters the number of historical values to display in the output. This allows to simplify the access to historical data.
A new section about model information is also added to the output. Now the trend, the cycles and the fluctuations are available.


Improvement and optimizations of existing services.

Release 1.6

Recommendation service

Based on historical transactional dataset of the form customer/item pairs, this service creates a recommendation model (recommender) and uses it to generate a list of items (recommendations) to suggest to a specific user.
Look at this video and/or this presentation which illustrate how to use the recommendation service.

Forecast service

This services now returns always the number of forecast requested.


  • Improvement and optimizations of existing services.

  • HTTP status 400 is returned when there is a syntax error in the call of a predictive services.

  • Available from all HCP landscapes:

Landscapes Hosts
 US East
US West

Release 1.5

Dev /ops improvements

Predictive service is more robust when several jobs are run concurrently.

Services improvements

Dataset service

Statistics information were computed during dataset registration. Now this is done for each variable of the dataset when you will get information about a specific variable. Statistics you will get are:
Advantages of this change:

  • Registration of dataset is faster

  • Statistics returned are computed based on the current state of the dataset.

Outliers service

When the input parameter to indicate the number of desired outliers is set to 0, the service will return all the outliers and 100 outliers are returned if no value is provided.

Forecast service

A new output parameter called “Maximum confident horizon” is provided. It represents the maximum number of forecast where the performance indicators of the best model are reliable.
Provide the maximum confident horizon in the output

The maximum confident horizon is added to the output
Time component of a date

When the date column contains time information, it appears also in the forecasts results.

How to get this new release?

To deploy it on your SAP Cloud Platform account,

  1. Go to the Services menu and click on the Predictive service tile

  2. In the page of the predictive service, click on the link: Go to Service

  3. The cockpit of the predictive service opens and alert you that a new version is available as shown in the screenshot. Click OK then click on the tile and deploy the new release.