30.03.2017 Views

company

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

SAP HANA supports this type of sidecar deployment<br />

to offer data scientists the ultimate flexibility<br />

to use any algorithm they want or even create<br />

their own in R. They call R scripts through<br />

SQLScript in SAP HANA and then pass the scripts<br />

to an external R server, along with all of the<br />

required input data. The results are sent back to<br />

SAP HANA and combined with any local data<br />

required to complete the query (see Figure 12).<br />

However, integrating the results from arbitrary<br />

scripting code using an external server results in<br />

significant overhead required to extract and<br />

transfer data to the R server for processing. Performing<br />

all calculations with a local in-memory<br />

predictive engine, such as the PAL for SAP HANA,<br />

avoids this overhead.<br />

PREDICTIVE ANALYSIS LIBRARY: PAL<br />

The predictive analysis library, or PAL, is designed<br />

to take advantage of the ability of SAP HANA to<br />

host execution engines and perform local calculations<br />

in memory. Unlike the previous option that<br />

uses an external predictive server for processing,<br />

this SAP HANA–native library enables users to<br />

perform in-database data mining and statistical<br />

calculations with excellent performance on large<br />

data sets.<br />

Figure 12: Sample Code in SQLScript<br />

/*Create a function using input and output<br />

table types in HANA*/<br />

CREATE FUNCTION LR( IN input1 SUCC_PREC_TYPE,<br />

OUT<br />

output0 R_COEF_TYPE)<br />

LANGUAGE RLANG AS’’’<br />

CHANGE_FREQ

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!