Advanced Data Science Using
SAS
Day 10 – Sun, 10 Dec 2024
Lal Bahadur Shastri Institute of
Management
G Krishnamurthy
Programing Constructs In SAS –
Using Arrays
Topics
• Recap
– Multiple Observations Per Subject
• Working With Arrays
– What is an Array
– Types of Arrays
– Performing An Operation on Every Variable
– Conversions From One Observation To Many Observations
– Conversions From Many Observation To One Observation
• Exercises
Recap
Working with Multiple
Observations in SAS
Introduction
• SAS normally processes one observation at a time
• However, in medical trials, longitudinal data
needs to be processes, that is, taking recordings
of people at different points of time
• SAS data sets for medical trials have time-based
monitoring of medical parameters
• Here, we discuss how SAS handles this kind of
data
Tools To Work with Longitudinal Data
• There are three types of programming tools
– LAG and DIF
• Using DIF to see the difference between FIRST and LAST
– First and Last variables
• Using SET-BY, followed by PUT
– Retained variables
• Used to compute the difference between first and last
record
Recap Exercises
• Please write the code to identify the purpose
of the lag function
• Consider the given program Exercise 2
– I have written a program, Exercise 2B
– Please explain why Exercise 2B is not working
Working With Arrays
Definition of Arrays
Need For Arrays
• A SAS array is nothing but a collection of SAS variables
• Using the array name and a subscript, an array element can
be used to represent any variable included in the array
• Arrays are created using ARRAY
• Array names are the same as variable names, along with a
bracket.
• You can use (), {}, or [] in defining an array name.
• Miss[1] represents the first variable in an array
• Consider a non-array program, Program 1, and we will
convert it to an array
Describing Numeric Arrays
• In the program, we use a DO-END loop to
process the data
– If Miss[i] = 999 then Miss[i] = .;
Describing Character Arrays
• Character arrays are created in the same
manner as numeric arrays
• We are using character arrays in Program 3 to
convert all values into upper case
• Please try this out for the SASHELP data set,
BASEBALL
Conversions – One Observation To Multiple
Observations
• Write a program to create a dataset wide that
lists one observation record for each subject
– Being done in Program 6
• Now, we need to convert this to a dataset that
has five observation records for each subject
– Being done in Program 7
• The reverse procedure has been done in
Program 8
Thank You
