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.
cancel
Showing results for 
Search instead for 
Did you mean: 

Here in this blog I have tried to consolidate all the information regarding SAP Predictive Analysis under one umbrella.  Main aim is to bring and assimilate the information relevant for SAP predictive analysis, say from system setup, to executing predictive algorithms, even for the beginners. I have also tried to retrieve information from other blogs as well.

SAP Predictive Analysis falls namely under two categories: Predictive Analysis Library and SAP Infinite Insight.


We can make use of Predictive Analysis Library (PAL) in mainly two ways:

  • Using HANA PAL libraries directly from HANA studio or
  • Using SAP Lumira Predictive Analysis Tool

HANA PAL

This is where it started. Once you get access to a HANA system (I hope you already have HANA studio in your system, if not please install HANA Studio), you cannot directly start working on PAL algorithms, there are certain prerequisites which you have to do to check whether HANA system is capable of executing PAL algorithms.

PAL Libraries are available from SAP HANA SPS06 onwards, but you can always go for the latest one, if it is available. With every upgrade, HANA team have tried to bring in lot of updates and features. HANA SPS08 has around 50+ PAL algorithms available. Basically PAL defines functions that can be called from within SQL Script procedures to perform analytic algorithms.

One can check if PAL libraries are successfully installed in your system or not by executing the following SQL statements in the SQL console.

SELECT * FROM "SYS"."AFL_AREAS" WHERE AREA_NAME = 'AFLPAL';

SELECT * FROM "SYS"."AFL_PACKAGES" WHERE AREA_NAME = 'AFLPAL';

SELECT * FROM "SYS"."AFL_FUNCTIONS" WHERE AREA_NAME = 'AFLPAL';

You will not see any results if PAL libraries are not installed in the HANA system which you are working on. You can contact your system administrator if the libraries are not installed or if you have system administrator access, you can follow the steps mentioned in this blog PAL Libraries setting up on HANA System.

Once this is done, then you need to give privileges to your user for executing PAL library functions. This can be done by executing the following statement:

GRANT EXECUTE ON system.afl_wrapper_generator to I068235;

GRANT EXECUTE ON system.afl_wrapper_eraser to I068235;

Here I have given my user name, but you cannot grant this privilege by logging in with your user (This is very important, you cannot give any kind of privileges to your own user after logging in with your user, always try to give privileges from a different user).


Once this is also done you are good to go. :smile:

You can check the SAP PAL Documentation for detailed description on PAL Alogrithms. All the PAL algorithms are explained with use case in this document.

This link always refers to the latest document regarding SAP HANA Predictive Analysis Library, and it will include the features of latest productive version of HANA.

You can see the examples of all the algorithms in the above mentioned document. It is very well explained. Only thing is that one has to select the best possible algorithm based on your use case and scenario. PAL Libraries/Algorithms are divided into 9 data mining categories. One frequently used category is Time Series algorithms. If you want to forecast any new values, these are the best algorithms available. There are five different algorithms under Time Series category.

E.g.: Double exponential smoothing.

You can watch the video on Double Exponential Time Series to have a better understanding about time series algorithms. This video has clearly explained the steps which you have to follow when you are working on PAL Time series algorithms. Similarly you can see the videos for other time series algorithms as well.

SAP Predictive Analysis and SAP Lumira

To avoid any confusion, SAP predictive analysis tool is altogether a different installation from SAP Lumira. If you have already installed SAP Lumira, you will have to uninstall SAP Lumira to install SAP predictive analysis.

You can download SAP predictive analysis from Service Market Place.

You need special privileges to download any software from service market place, normally most people don’t have it. You can ask for permission from the same page itself, it will go to your direct reporting manager for approval.

Once you install the tool, it will be a 30 day trial, license will get expired after 30 days.

You can watch the video on SAP predictive analysis tool setup to have an idea about installation and setting up of SAP predictive analysis tool.

In this video they have covered connecting to a HANA system from Predictive Analysis tool as well. While trying to connect to the HANA systems, try to give SAP HANA server as: lddb<system ID>.wdf.sap.corp

Once it is connected you can directly pull data from the tables as mentioned in the video.

If you have already tried out Double Exponential Time series from HANA studio, the next steps will be easy.

You can drag and drop the algorithms which you want in the predict tab. The screen will look like this now.

Once this is done, you can change the properties of the algorithm by clicking on the settings button on the selected icon. Then select configure settings, and then a screen will come where in you have to enter all the mandatory values. See the screen shot below to see an example.

Once the configuration is done, run the algorithm from the same screen. You will see the results in a table, and if you go to Trend chart you will be to see the predicted values in a graph like below screen.

Working with this tool is this easy, and once we get the results from the algorithm you have the option to write the data back to the HANA DB as well.

Most of the PAL algorithms available in HANA systems are available in SAP Predictive Analysis Tool except few. Selecting a particular algorithm is as easy as drag and drop. You don’t even have the additional overhead of creating signature tables for calling a PAL algorithm. (You will come across signature tables if you try to call any PAL algorithm from HANA Studio, we even have to create a result table which will store the result data once we successfully execute the algorithm). Here in this tool, everything can be maintained as properties for whatever algorithm which you have selected. Once you execute the algorithm the results can be displayed directly on a graph.

SAP Infinite Insight

SAP has bought KXEN to mainly deal with automated predictive analysis. Now with this acquisition SAP has renamed the software to SAP infinite Insight.

Infinite Insight can be downloaded from Service Market Place.

There are different versions available in service market place. Latest version will occupy 2.5 GB in space, we don’t have to install the entire setup, to make it easier one can download the object ‘IIWS7000_0-80000274.EXE'. You can give this as search term and download the file. Image below is the screenshot for searching the same so that .exe file comes in the search results.

Once you install SAP Infinite Insight, you can directly start working on it. Once you open the software you will see a screen like this:

SAP infinite insight is a vast topic and there are lots of features associated with it.

SAP Infinite insight help on SDN will give you a fair idea about the tool and the features that its offering.

If you have any doubt regarding setting up of the SAP Infinite Insight Tool and connecting to a particular DBMS you can go to the SAP Infinite Insight Help Portal.

It gives in depth understanding of each and every topic and all the features like Explorer, Modeler, Social, Recommendation and Tool kit are explained in detail in separate documents.

There is already some interesting blogs written for Explorer and Modeler. You can read that as well.

Since it is very difficult to cover all the features in one single blog, I will try to write another blog exclusively for SAP infinite Insight considering one use case covering E2E functionalities.

Feedbacks are welcome :smile:

6 Comments