High-Performance Aggregation in SAS Viya: Mastering CAS GroupBy Actions

Difficulty Level

Beginner

Published on : 11/10/2021

Expert Advice

Stéphanie
Spécialiste Machine Learning et IA.

The final step—saving as .sashdat—is what makes this workflow "production-ready." .sashdat files are not just data; they are the memory-image of the CAS table. When you reload this file later, CAS doesn't have to "read" the data in the traditional sense; it simply maps the file into memory, providing near-instant access.

Unleashing the Power of CAS Actions

Grouping data in SAS Viya's CAS environment is fundamentally different from traditional Base SAS. Instead of procedures, we use CAS Actions, which are the building blocks of the in-memory engine.

This example demonstrates how to create a group-by table using CAS (Cloud Analytic Services) actions on the SAS^© Viya^™ platform. The input table CreditQualify is grouped by State variables and State FIPS codes. The Credit_Score variable is used as a weighting variable to calculate the mean (MEAN) of credit scores per group. Intermediate and final results are managed using CAS actions such as simple.groupBy, table.alterTable, table.fetch, and table.save. Once grouping is complete, the output table ScorePerState is enhanced with a formatted _Score_ column and saved in .sashdat format for easy reuse. The severity code check (s.severity = 0) ensures that operations are executed only upon successful completion of the previous action.

Data Analysis

Type : CREATION_INTERNE

The example CreditQualify table is created using a DATA step with datalines to ensure the example's autonomy. The data is fictitious and includes the variables State (character string), State_FIPS (numeric), and Credit_Score (numeric).

1 Code Block

PROC CAS / DATA STEP Data

Explanation :
The initial DATA step creates a temporary table named 'CreditQualify' with example data for the 'State' (state), 'State_FIPS' (state FIPS code), and 'Credit_Score' (credit score) variables. The PROC CAS statement activates the SAS Cloud Analytic Services (CAS) session. The simple.groupBy action is then used to build groups based on the 'State' and 'State_FIPS' variables. 'Credit_Score' is defined as the weighting variable, and the aggregator is set to 'MEAN' to calculate the average credit scores per group. The 'scoregt' and 'scorelt' options define the lower and upper limits of scores to include. The result of this action is an output table named 'ScorePerState' in the active caslib. A severity code check (s.severity = 0) ensures that subsequent actions are executed only upon success. If the groupBy action succeeds, the table.alterTable action is used to modify the '_Score_' column by assigning it the label 'Credit Score' and format '5.2'. Then, the table.fetch action retrieves the 'State', 'State_FIPS', and '_Score_' variables from the 'ScorePerState' table, applying the specified format for '_Score_'. Finally, the table.save action saves the 'ScorePerState' table in .sashdat format in the active caslib.

Copied!

1	DATA work.CreditQualify;
2	INPUT State $ State_FIPS Credit_Score;
3	DATALINES;
4	NC 37 700
5	NC 37 750
6	NY 36 600
7	NY 36 620
8	CA 06 800
9	CA 06 820
10	;
11	RUN;
12
13	PROC CAS;
14	SESSION mysession;
15	SIMPLE.groupBy RESULT=r STATUS=s /
16	inputs={"State", "State_FIPS"},
17	weight="Credit_Score",
18	aggregator="MEAN",
19	scoregt=0,
20	scorelt=900,
21	TABLE={name="CreditQualify"},
22	casout={name="ScorePerState",
23	replace=true};
24	RUN;
25	IF (s.severity = 0) THEN DO;
26	TABLE.alterTable / columns={
27	{label="Credit Score", FORMAT="5.2", name="_Score_"}},
28	name="ScorePerState";
29	TABLE.fetch /
30	FORMAT=True,
31	fetchVars={"State", "State_FIPS",
32	{name="_Score_",FORMAT="5.2"}},
33	TABLE={name="ScorePerState"},
34	index=false;
35
36	TABLE.save /
37	TABLE={name="ScorePerState"},
38	name="ScorePerState.sashdat",
39	replace=True;
40	END;
41	RUN;
42	QUIT;

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.