Skip to Content

By Vishwanath Belur and Clarissa Dold

Here is a preview of some of the planned new features in the soon to be released Predictive Analysis SP11.

1) Tighter integration with SAP Lumira workflows

You can now save Predictive Analysis datasets and custom PA visualizations as .svid. When opening the .svid file your datasets and
visualizations will automatically be re-created.

1_Lumira.jpg

2_Lumira.jpg

2) Ability to extend the algorithm set of Predictive Analysis with new algorithms

With SP11 you can customize Predictive Analysis with your own R algorithms. You can run any R-script (HANA or non-HANA)
from the PA console (generic R-support) allowing you to add your own components to Predictive Analysis.

3_OwnR.jpg

and configure the properties of that component.

4_OwnR.jpg

Example: Custom R – Results and R visualizations in Predictive Analysis

11_OwnR.jpg

3) HANA SPS5 PAL Parity – support of new algorithms and data preparation

Algorithms

    • Regression: Logistic Regression
    • Outlier Detection: Variance Test

5a_PALSupport.jpg

Data Preparation

    • Binning
    • Scaling Range (Normalization)

5b_PALSupport.jpg

4) Enabling of Model Consumption

As of SP11 you can convert the HANA PAL models created in Predictive Analysis into a SQL procedure for SAP HANA consumption.

8_ExportPAL.jpgx

9_ExportPAL.jpg

In the HANA Studio you will see the following:

10_ExportPAL.jpg

To report this post you need to login first.

9 Comments

You must be Logged on to comment or reply to a post.

  1. Star buck

    About the R-integration:

    I notice there is a ‘load script’ button, so I suppose it’s possible to create R-scripts in any tool and then load it into the PA console?

    PS. is the R-version still 2.15.x?

    Nice stuff BTW.

    (0) 
    1. Vishwanath Belur

      Hello,

      Yes, you can bring in R scripts that you have written anywhere, just that they need to be in function format.

      Yes, R revision is still 2.15.

      Thanks,

      Vishwa

      (0) 
  2. Balamuralikrishna S

    I am working with PA 1.0.8 . In this version under Regression I can see Exponential Regression , Geometric Regression , Linear Regression same way R-Exponential Regression , R-Geometric Regression , R-Linear Regression. I believe the algorithms prefixed with R uses R default Algorithms. Similarly I can see in the screenshots above there is HANA Apriori and HANA R-Apriori. When it comes to usage of these two kinds what will be the difference in results ?

    (0) 
    1. Henrique Pinto

      There are 4 kinds of algorithms in PA:

      – native algorithms: which logic is implemented natively in PA’s core, names are like <algo name>

      – desktop R algorithms: implemented with R scripts that are run against a local (desktop) R installation, names are like R-<algo name>

      – HANA algorithms: implemented on top of PAL (Predictive Analysis Library) functions, they run in-database in HANA, algos names are like HANA-<algo name>

      – HANA-R algorithms: implemented as R-scripts which are run against a R server connected to a HANA box, algos names are like HANA R-<algo name>

      Best,

      Henrique.

      (0) 

Leave a Reply