*Copyright @ www.mycsg.in;

What is the word meaning of subset

1. A part of a larger group of related things.

Subset observations from a dataset

Where do we need to subset data

• Subset the records of cars manufactured in a particular continent
• Select only the male student records from the class dataset
• Identify and list the subjects of a clinical trial whose age is greater than 60

Create sample input datasets

data class; set sashelp.class; run; data cars; set sashelp.cars; run; Copy Code View Log

• `class` and `cars` will be used in the subsetting examples below
• Inspect the input datasets first, so you can clearly see which rows are retained by each subsetting technique

View Data

Subset records using the WHERE statement

• The `where` statement filters observations based on a condition
• Only observations that meet the condition are read from the input dataset into the DATA step processing stream
• This can be more efficient than reading all observations and discarding some later

Create a dataset named males by subsetting the male student records from class dataset

• Notice in the log the number of observations read and written
• Only observations with `sex="M"` are selected

data males; set class; where sex="M"; run; Copy Code View Log

• Inspect `males` and confirm that all retained observations are male students

View Data

Create a dataset named females_preteen by subsetting female student records whose age is less than 13

• Here, the filter depends on two variables, `sex` and `age`
• The conditions are combined with `and` because both must be true

data females_preteen; set class; where sex="F" and age lt 13; run; Copy Code View Log

• Verify that every retained observation is female and younger than 13

View Data

Create a dataset named cars_mileage by subsetting cars with mpg_city greater than 24 or mpg_highway greater than 30

• This example combines two conditions with `or` because a row is retained when at least one condition is true
• Such compound conditions are common in real filtering tasks

data cars_mileage; set cars; where mpg_city gt 24 or mpg_highway gt 30; run; Copy Code View Log

• Inspect the retained rows and confirm that each car satisfies at least one of the two mileage conditions

View Data

Subset records using a subsetting IF statement

• When a subsetting IF statement is used, SAS first reads the observation into the PDV and then evaluates the condition
• If the expression evaluates to false, SAS returns to the top of the DATA step without writing the observation
• This is a flexible approach because the IF condition can use variables created earlier in the same DATA step

Create a dataset named females by subsetting female records using a subsetting IF statement

data females; set class; if sex="F"; run; Copy Code View Log

• Compare `females` with the earlier WHERE example, and note that the final selected rows can be the same even though processing differs internally

View Data

Create a dataset named select_sedans by subsetting Sedan cars with highway mileage greater than 35 using subsetting IF

data select_sedans; set cars; if type="Sedan" and mpg_highway gt 35; run; Copy Code View Log View Data

Create a dataset named cheap_mileage by subsetting cars with msrp less than 15000 or city mileage greater than 40 using subsetting IF

data cheap_mileage; set cars; if msrp lt 15000 or mpg_city gt 40; run; Copy Code View Log View Data

Subset records using IF THEN OUTPUT

• The `output` statement tells SAS to write the current contents of the PDV to the output dataset
• With `if condition then output;`, we explicitly write only rows that satisfy the condition
• This pattern is useful when one DATA step may write different observations to different output datasets

Create a dataset named tall by subsetting students with height greater than 60 using IF THEN OUTPUT

data tall; set class; if height gt 60 then output; run; Copy Code View Log

• Inspect `tall` and confirm that every student has height greater than 60

View Data

Subset records using IF THEN DELETE

• The `delete` statement tells SAS to discard the current observation and return to the top of the DATA step
• With `if condition then delete;`, we remove unwanted rows instead of explicitly writing wanted rows
• This is especially convenient when it is easier to describe which observations should be excluded

Create a dataset named short by subsetting students with height less than 60 using IF THEN DELETE

• Because we want the short students, we delete observations whose height is greater than or equal to 60

data short; set class; if height ge 60 then delete; run; Copy Code View Log

• Review `short` and confirm that only students with height less than 60 remain

View Data

Key points to remember

• `where` filters observations before they are processed in the DATA step
• A subsetting `if` filters after the observation has been read into the PDV
• `if then output` explicitly writes selected observations
• `if then delete` explicitly discards selected observations
• Different subsetting styles can produce the same final result, but differ in processing behavior and flexibility