Starting with KXEN - [Updated with more info]
Apologies I marked this a question, I would like this to be a part of discussion and update this as and when we know more about KXEN:
Would KXEN be the future predictive engine for SAP; Is it an alternate for SAP PA with more robust technology; SAP PA is more recommended for HANA users; As HANA users, we are using SAP PA; PA as such is very easy to use tool and it solves our purpose in Analytics and Visual Intelligence.
[like SAP has BODS and BW- 2 ETL section for different section of customers- though BODS would be able to cover most of BW capabilities-didn want to divert the topic to ETL side];
In one of the post, it was mentioned that SAP was using SPSS as engine and once IBM bought it, SAP moved towards R, which is an open source. now according to the expert SAP PA's cost equation was not balanced as they used open source engine; and so SAP bought KXEN that was in "strong performer's" quadrant;
So now, all said and done SAP has PA & KXEN; any idea how & where all KXEN and PA fit the SAP analytics landscape/ architecture; and How do we start with KXEN, is there a trial version for us to start with. I am excited and eager to start with KXEN;
More information on KXEN-
KXEN has a broad split of tools with Explorer, Modeler & Social network analytics. ... well this looks usual but it is not ...
Social network analysis is a major factor that differentiates KXNE from other tool; [majorly big data] all 400 major clients: other target areas are Turnover, cross sell, upsell, customer experience, fraud analysis & Risk analysis;
Infinite insight is one of the tool used by KXEN, designed as a statistical data mining tool & their algorithms are integrated & Optimized for efficient processing of BIG data;
Other stat packages including SAS, SPSS, SAP PA have libraries of algorithms developed by various contributors focused on maximum likely hood estimation matrix and require highly trained analyst programmers.
It is said that, KXEN product INFINITE INSIGHT build model very fast allowing analyst to leverage their expertise in understanding relationship and tuning models rather than tedious data prep and variable screening:
To explain more ...
TRADITIONAL PROCESS ...
Select variable - reduce 30 variables from 100 that the tool could handle.
Data Preparation- 60% of the expert's project time [like removing outliers, remove missing data etc.]
Build model & Test- This is an iterative process [average of 3 weeks is anticipated to be aggressive for the above model]
Infinite insight patent Process proposes to shrink this model like this can be done in hours and reduce expert's time .. and of course $$; Apart from that, if integrated with HANA, there will be no dataset sampling problems ... variable selections problems... or processing time problems.
Their patented application is called "Structural Risk Minimization" [they call it SRM-patented by KXEN]- this automates data prep, variable encoding & Variable screening and model validation.
When it comes to output, it provides explanation of the output rather than just the NUMBERS in output; it tells you which contributes to the prediction and their %; confirms the model quality & model robustness. What that numbers mean and explanation of the numbers; It allows you to export executive summary to power point.
SAP Analytics Game changing strategy will be well complimented by KXEN ... with this experts could anticipate SAP along with KXEN would move to the top of leader's quadrant.
Message was edited by: Arvind Elayappan