Beyond the Basics: Using PARTITION and ALTERTABLE to Reshape CAS Data

Difficulty Level

Beginner

Published on : 15/10/2024

Expert Advice

Michael
Responsable de l'infrastructure Viya.

While the PARTITION statement in PROC CASUTIL is often associated with creating data partitions for sampling, it is also highly effective as a high-performance filtering tool (as shown in your code with where='...'); unlike a standard DATA step which processes row-by-row, this method performs the subsetting directly in the CAS server's memory, often resulting in faster execution for large datasets.

Use PROC CASUTIL for efficient, code-lite management of in-memory tables.

This documentation presents an example of using the PROC CASUTIL procedure to manipulate data in the SAS^© Cloud Analytic Services (CAS) environment. The example illustrates how to load an internal dataset (sashelp.cars), partition it by selecting observations based on specific criteria, and then modify the resulting table to keep only a subset of columns. It also shows how to print the resulting CAS table.

Data Analysis

Type : INTERNAL_CREATION

Examples use generated data (datalines) or SASHELP.

1 Code Block

PROC CASUTIL

Explanation :
The code begins by establishing a CAS session (`cas casauto`) and assigning a library (`libname mylib cas`) to access CAS tables.
The `proc casutil;` statement initializes the CASUTIL procedure.
1. The `load data=sashelp.cars casout='cars' replace;` statement loads the `sashelp.cars` dataset (an internal SAS dataset) into CAS memory under the name 'cars'. The `replace` option ensures that the table is replaced if it already exists.
2. The `partition casdata='cars' casout='carsWhere' replace where='MSRP>90000 and Make="Porsche"';` statement creates a new CAS table named 'carsWhere'. It is created by selecting rows from the 'cars' table where the 'MSRP' value is greater than 90,000 and 'Make' is "Porsche".
3. The `altertable casdata="carsWhere" keep={"make", "model", "MSRP"};` statement modifies the 'carsWhere' table to keep only the 'make', 'model', and 'MSRP' columns.
Finally, `proc print data=mylib.carsWhere;` displays the content of the modified 'carsWhere' CAS table.

Copied!

1	cas casauto sessopts=(caslib='casuser');
2	LIBNAME mylib cas;
3
4	PROC CASUTIL;
5	load DATA=sashelp.cars
6	casout='cars' replace;
7	partition casdata='cars'
8	casout='carsWhere' replace
9	where='MSRP>90000 and Make="Porsche"';
10	altertable casdata="carsWhere"
11	keep={"make", "model", "MSRP"};
12	QUIT;
13	PROC PRINT DATA=mylib.carsWhere;

Pro Tip

Perform your column selection (KEEP) and row filtering (WHERE) as early as possible—ideally during the load or partition phase—to minimize the memory footprint of your CAS session.

This material is provided "as is" by We Are Cas. There are no warranties, expressed or implied, as to merchantability or fitness for a particular purpose regarding the materials or code contained herein. We Are Cas is not responsible for errors in this material as it now exists or will exist, nor does We Are Cas provide technical support for it.