Appendix C: SAS Software

1
Appendix C SAS Software
Uses of SAS

• linear programming
• forecasting
• econometrics
• nonlinear parameter estimation

• CRM
• datamining
• data warehousing

• simulation
• marketing models
• statistical analysis

Data Types SAS Can Deal with

• Web log data
• questionnaires

• panel data
• relational databases

• scanner data

Ideal When You Are

• massaging
• sorting
• merging

• lookups
• reporting

• transforming
• manipulating

2
Two Types of SAS Routines

• DATA Steps
• Read and Write Data
• Create a SAS dataset
• Manipulate and Transform Data
• Open-Ended - Procedural Language
• Presence of INPUT statement creates a Loop
• PROC Steps
• Analyze Data
• Canned or Preprogrammed Input and Output

3
A Simple Example

• data my_study
• input id gender green recycle
• cards
• 001 m 4 2
• 002 m 3 1
• 003 f 3 2
• proc reg datamy_study
• class gender
• model recycle green gender

4
The Sequence Depends on the Need

• data step to read in scanner data
• data step to read in panel data
• data step to merge scanner and panel records
• data step to change the level of analysis to the
  household
• proc step to create covariance matrix
• data step to write covariance matrix in LISREL
  compatable format

5
The INPUT Statement - Character Data

• List input
• after a variable - character var
• input last_name first_name initial
• Formatted input
• w. after a variable
• input last_name 22. first_name 22. initial
  1.
• Column input
• start-column - end-column
• input last_name 1 - 22 first_name 23 - 45
  initial 46

6
The INPUT Statement - Numeric Data

• List input
• input score_1 score_2 score_3
• Formatted input
• w.d (field width and number of digits after an
  implied decimal point) after a variable
• input score_1 10. score_2 10. score_3 10.
• Column input
• start-column - end-column
• input score_1 1 - 10 score_2 11 - 20 score_3 21
  - 30

7
Grouped INPUT Statements

• input (var1-var3) (10. 10. 10.)
• input (var1-var3) (310.)
• input (var1-var3) (10.)
• input (name var1-var3) (10. 35.1)

8
The Column Pointer in the INPUT Statement

• input _at_3 var1 10.
• input more _at_
• if more then input _at_15 x1 x2
• input _at_12 x1 5. 3 x2

9
Documenting INPUT Statements

• input _at_4 green1 4. / greeness scale
  first item /
• _at_9 green2 4. / greeness scale 2nd
  item /
• _at_20 aware1 5. / awareness scale
  first item /
• _at_20 aware2 5. / awareness scale
  2nd item /

10
The Line Pointer

• input x1 x2 x3 / x4 x4 x6
• input x1 x2 x3 2 x4 x5 x6
• input x1 x2 x3
• 2 x4 x5 x6

11
Reading an External File on Unix

• data
• filename raw_sem 'my_garnet_disk_file.data'
  
• infile raw_sem
• input a b etc.

12
The PUT Statement

• put _all_
• put a b
• put _infile_
• put _page_
• col1 22 col2 14
• put _at_col1 var245 _at_col2 var246

• put x1 x2 x3 _at_
• input x4
• put x4
• put x1 2 x2
• put x1 / x2

13
Copying Raw Data

• data _null_
• infile in
• outfile out
• input
• put _infile_

14
SAS Constants

• '21Dec1981'D
• 'Charles F. Hofacker'
• 492992.1223

15
Assignment Statement

• x a b
• y x / 2.
• prob 1 - exp(-z2/2)

16
The SAS Array Statement

• array y 20 y1-y20
• do i 1 to 20
• yi 11 - yi
• end

17
The Sum Statement

• variableexpression

retain variable variable variable
expression
n1 cumulated x
18
IF Statement

• if a gt 45 then a 45
• if 0 lt age lt 1 then age 1
• if a 2 or b 3 then c 1
• if a 2 and b 3 then c 1
• if major "FIN"
• if major "FIN" then do
• a 1
• b 2
• end

19
More IF Statement Expressions

• name ne 'smith'
• name 'smith'
• x eq 1 or x eq 2
• x1 x2
• a lt b a gt c
• a le b or a le c
• a1 and a2 or a3
• (a1 and a2) or a3

if
then etc
20
Concatenating Datasets Sequentially
first id x y 1 2 3 2 1 2 3 3 1
second id x y 4 3 2 5 2 1 6 1 1

• data both
• set first second

both id x y 1 2 3 2 1 2 3 3 1 4 3 2 5 2
1 6 1 1
21
Interleaving Two Datasets

• proc sort datastore1
• by date
• proc sort datastore2
• by date
• data both
• set store1 store2
• by date

22
Concatenating Datasets Horizontally
left id y1 y2 1 2 3 2 1 2 3 3 1
right id x1 x2 1 3 2 2 2 1 3 1 1

• data both
• merge left right

both id y1 y2 x1 x2 1 2 3 3 2 2 1 2 2
1 3 3 1 1 1
23
Table LookUp
table part desc 0011 hammer 0012 nail 0013
bow
database id part 1 0011 2 0011 3 0013

• proc sort datadatabase outsorted
• by part
• data both
• merge table sorted
• by part

both id part desc 1 0011 hammer 2 0011
hammer 3 0013 bow
24
Changing the Level of Analysis 1

• Day Score Student
• 1 12 A
• 1 11 B
• 1 13 C
• 2 14 A
• 2 10 B
• 2 9 C
• Day Highest Student
• 1 13 C
• 2 14 A

Before
After
25
Changing the Level of Analysis 1FIRST. and LAST.
Variable Modifiers

• proc sort datalog
• by day
• data find_highest
• retain hightest
• drop score
• set log
• by day
• if first.day then highest.
• if score gt highest then highest score
• if lastday then output

26
Changing the Level of Analysis 2

• Subject Time Score
• A 1 A1
• A 2 A2
• A 3 A3
• B 1 B1
• B 2 B2
• B 3 B3
• Subject Score1 Score2 Score3
• A A1 A2 A3
• B B1 B2 B3

Before
After
27
Changing the Level of Analysis 2

• data after
• keep subject score1 score2 score3
• retain score1 score2
• set before
• if time1 then score1 score
• else if time2 then score2 score
• else if time3 then do
• score3 score
• output
• end

28
The KEEP and DROP Statements

• keep a b f h
• drop x1-x99
• data a(keep a1 a2) b(keep b1 b2)
• set x
• if blah then output a
• else output b

29
Changing the Level of Analysis 3Spreading Out an
Observation

• Subject Score1 Score2 Score3
• A A1 A2 A3
• B B1 B2 B3
• Subject Time Score
• A 1 A1
• A 2 A2
• A 3 A3
• B 1 B1
• B 2 B2
• B 3 B3

Before
After
30
Code for Change 3

• data spread
• drop score1 score2 score3
• set tight
• time 1 score score1 output
• time 2 score score2 output
• time 3 score score3 output
  

31
Use of the IN Dataset Indicator

• data new
• set old1 (infrom_old1)
• old2 (infrom_old2)
• if from_old1 then
• if from_old2 then

32
Proc Summary for Aggregation

• proc summary dataraw_purchases
• by household
• class brand
• output outhousehold countx xy
• VAR variable(s)lt/ WEIGHTweight-variablegt

33
Using SAS for Simulations

• data monte_carlo
• keep y1 - y4
• array y4 y1 - y4
• array loading4 l1 - l4
• array unique4 u1 - u4
• l1 1 l2 .5 l3 .5 l4 .5
• u1 .2 u2 .2 u3 .2 u4 .2
• do subject 1 to 100
• eta rannor(1921)
• do j 1 to 4
• yj etaloadingj
  uniquejrannor(2917)
• end
• output
• end
• proc calis datamonte_carlo
• etc.

Simulation Loop