Introducing the Automated Predictive Library (APL) for SAC AND HANA
Where does APL come from?
When SAP acquired KXEN in 2013, the company offered a software solution to help companies to create and manage predictive models in a semi-automated mode. This engine has evolved to meet SAP’s quality criteria and has been adapted to run in HANA. It then took the name Automated Predictive Library (APL). It is constantly evolving and improving, for example by integrating new Machine Learning algorithms, while respecting what makes it unique, i.e. being an automated Machine Learning engine.
Today, APL provides Machine Learning algorithms for SAP HANA, and SAP Analytics Cloud. It is also used in a growing number of SAP applications and solutions, and as such has become an important part of the SAP Predictive and Machine Learning landscape.
What is APL?
SAP Automated Predictive Library (APL) is a HANA Application Function Library. Technically APL embeds a pre-packaged predictive analytics engine in SAP HANA. The objective of APL is to make the Automated Analytics technology available inside HANA, and make it available to any application that’s able to use and consume HANA services, be it on premise or on the cloud.
Which SAP solutions and applications use APL?
- APL is provided in standard with HANA 2.0 on premise SPS04 and later versions.
- APL is pre-installed in HANA Cloud. You must enable the script server.
SAP Analytics Cloud
- APL is embedded into SAP Analytics Cloud as the machine learning engine that enables Smart Assist to transparently deliver augmented performance metrics.
- Smart Predict delegates its predictive modeling functions to APL when the data source is HANA on premise (Live Predict).
- SAP HANA 1.0 SPS 12 Revision 122.04 and above, SAP HANA 2.0 SPS04 Revision 47 and above, and SAP HANA 2.0 SPS05 Revision 51 and above are supported.
Other SAP solutions
- BW developers can use APL. They have the ability to create objects in HANA by developing ABAP/BW design artifacts using Eclipse.
- SAP Data Intelligence can train and save APL predictive models in HANA thanks to “ML training” operator and “Artifact Producer” object.
- Thanks to the HANA ML Python API developers can can benefit from APL in SAP Data Warehouse Cloud.
- Intelligent Scenario Lifecycle Management (ISLM) is the solution for those who need to embed APL in S/4HANA or in a SAP Business Application.
How can SAP customers leverage APL?
Customers use the functions provided by APL to implement a predictive modeling process to answer traditional business questions on their data in HANA. Automated predictive modelling capabilities are available to any application that can use and consume HANA services, on premise or in the cloud.
APL is available for HANA Enterprise Edition, SAP HANA Predictive Option (can be purchased as part of the HANA Standard, and SAP HANA Cloud.
Why is APL important to SAP Analytics Cloud?
APL is the predictive engine used by Smart Discovery, Smart Predict, and Predictive Planning.
No installation is required. It provides automatically augmented insights to all SAP Analytics Cloud users. They do not need to know APL, its use is completely transparent to them.
APL is also the predictive engine used by Smart Predict on Live HANA. In that case, Smart Predict delegates processing to APL in HANA on-premise, so that Machine Learning process run the closest to the data. HANA on premise customers need to download and install APL on HANA to make Smart Predict work correctly in live mode.
Customers with HANA Runtime
- Customers With HANA Runtime have the right to use APL only through the application attached to their HANA Runtime license.
- BW customers with a HANA Runtime license cannot use Smart Predict on Live HANA because it does not yet support calculation views, and it is not allowed to create and use tables & SQL views.
Important warning: The SAC functionality (Live Predict) for building and executing SAC predictive workflows directly in HANA Cloud is not supported yet.
APL is a key mandatory component of SAP Analytics Cloud. It is provided as standard with SAC because it is necessary to operate Smart Assist. It must be installed in HANA to enable Live Predict and delegate SAC Smart Predict analysis directly in HANA. To date, Live Predict is only available with HANA on premise.
APL is also a very useful library for HANA developers who want to build predictive models in a simple way and embed them in SQL, C++, Java, etc in operational workflows.