copyright © 2006, sas institute inc. all rights reserved. a sampler of what's new in base sas...

Copyright © 2006, SAS Institute Inc. All rights reserved.

A Sampler of What's New in Base SAS [email protected]


Caramel, Nut, or Truffle?

DATA Step• User-Written Functions

• Data Set Lists

• Current Input Data Set Name

• INPUT with String Delimiter

PROC SQL• Un-PUT

• Faster DISTINCT

PROC SORT Linguistic Ordering

Encrypted Macro Storage


Wrap in a Macro

%macro study_day(start, event, study_day);

n = &event - &start;

if n >= 0 then

&study_day = n + 1;

else

&study_day = n;

%mend;


Bug in the Macro

data results;

set trial_data;

n = sum(of visits_jan--visits_dec);

%study_day(start, event, study_day)

if n > 5 then

...


Bug in the Macro

data results;

set trial_data;


n = event – start;

if n >= 0 then

study_day = n + 1;

else

study_day = n;

if n > 5 then

...


Protected by a Function

data results;

set trial_data;


study_day = study_day(start, event);

if n > 5 then

...


Study Day Function

proc fcmp outlib=sasuser.funcs.trial;

function study_day(start, event);

n = event – start;

if n >= 0 then

return(n + 1);

else

return(n);

endsub;


Data Set Lists

#1 on 2006 SASware Ballot

Read Like-Named Data Sets

Syntax borrowed from DROP and KEEP


Read a Bunch of Data Sets

data out;

set a1 a2 a3 a4 a5 a6 a7 a8 a9 a10

a11 a12 a13 a14 a15 a16 a17 a18

a19 a20 a21 a22 a23 a24 a25 a26

a27 a28 a29 a30 a31 a32 a33 a34

a35 a36 a37 a38 a39 a40 a41 a42

a43 a44 a45 a46 a47 a48 a49 a50;

run;


Data Set Lists

data out;

set a1-a50(pw=pass123);

run;

data out;

set sales.jan:;

run;


String Delimiter

"Top 10" SASware Ballot Item

Difficult to parse multi-character delimiters

INFILE DLMSTR=

Works like DLM=


String Delimiter

data bad_guys;

length date time ip $ 16;

infile datalines dlm=' ' dsd;

input date time ip;

datalines;

"20APR2007" "12:56:46" "146.16.1.23"

run;


String Delimiter

data bad_guys;

length date time ip $ 16;

infile datalines dlmstr=' ' dsd;

input date time ip;

datalines;

"20APR2007" "12:56:46" "146.16.1.23"

run;


SET with INDSNAME=

"Top 10" SASware Ballot Item

Place current data set name into a variable

SET with INDSNAME = var


SET with INDSNAME=

data results;

set gas_price_option

gas_rbid_option

coal_price_forward

coal_rbid_forward

indsname = cur_dataset;

...

run;


DATA Step JavaObj

"Compute" methods written in Java

Call from DATA Step with JavaObj

Methods like Java Native Interface (JNI)

Uses dot syntax

Production in SAS 9.2

Experimental in SAS 9.1.3


data results;

declare javaobj jo("myclass");

jo.callDoubleMethod("compute",

x, y, z, res);

...

DATA Step JavaObj


DATA Step IN Operator

IN operator used simple array search

Now, IN searches a binary tree

9 12 14 17 19 23 50 54 67 72 76


data _null_;

set customers;

if state in ('NC', 'SC', 'GA',

'TN', 'FL');

run;

DATA Step IN Operator


Proc SQL Today

proc format;

value udfmt 1-3='small'

4-6='medium'

7-9='large';

proc sql;

select style as SmallStyles

from oracle.clothes

where put(size, udfmt.) = 'small';


DBMS

Proc SQL Today

SAS® Session

select style as SmallStylesfrom oracle.clotheswhere put(size, udfmt.) = 'small';

SAS/Access Engine

select style,size from clothes


SAS® Session

select style as SmallStylesfrom oracle.clotheswhere put(size, udfmt.) = 'small';

DBMS

Proc SQL Tomorrow

SAS/Access Engine

select style as SmallStylesfrom oracle.clotheswhere (1 <= size and size <= 3);


Proc SQL Today

proc sql;

select name

from oracle.employees

where put(bday, date5.) = '31OCT';


Proc SQL Tomorrow

proc sql;

select name

from oracle.employees

where month(bday) = 10 and

day(bday) = 31;


Proc SQL Today

proc format;

value nudfmt 0 = 'RED'

1 = 'NOTRED'

2 = 'GREEN'

other = 'BLACK';

proc sql;

select * from oracle.data where

put(color, nudfmt.) contains "RED";


Proc SQL Tomorrow

proc format;

value nudfmt 0 = 'RED'

1 = 'NOTRED'

2 = 'GREEN'

other = 'BLACK';

proc sql;

select * from oracle.data where

color in (0,1);


PROC SQL

Faster: SELECT DISTINCT

Faster: COUNT(DISTINCT var)

Hash Join Performance

Row Counts from DBMSes


SELECT DISTINCT

proc sql;

create table distinct_values as

select distinct month, year

from customer_info;

quit;


COUNT DISTINCT

proc sql;

create table count_distinct as

select count(distinct customerid)

from customer_info;

quit;


Linguistic Sorting

Obs Binary Lowfirst Linguistic

1 Aaron aardvark aardvark

2 Aztec azimuth Aaron

3 Zeus Aaron azimuth

4 aardvark Aztec Aztec

5 azimuth zebra zebra

6 zebra Zeus Zeus


Linguistic Sorting

proc sort data=names

out=sorted

sortseq=linguistic;

by name;

run;


Numeric_Collation=ON

Obs Binary Linguistic

1 0123 Main St. Apt #12 123 Main St. Apt #1






Encrypted Macro Storage

%macro mymac(x, y) / store secure;

...

%mend;


More Details

Search support.sas.com for:

Papers:

"User-Written DATA Step Functions"

"Creating Order out of Character Chaos"

copyright © 2006, sas institute inc. all rights reserved. a sampler of what's new in base sas...

Documents