This tutorial explains how to use PROC FREQ with various examples. The PROC FREQ is one of the most frequently used SAS procedures which helps to summarize categorical variable. It calculates count/frequency and cumulative frequency of categories of a categorical variable. In other words, it returns the number and percentage of cases falling in multiple categories of a categorical variable. It's not just restricted to counts. It also produces bar charts and tests for association between two categorical variables.

**Create a sample data set**

The program below creates a sample SAS dataset which would be used in further examples to explain PROC FREQ.

data example1;

input x y $ z;

cards;

6 A 60

6 A 70

2 A 100

2 B 10

3 B 67

2 C 81

3 C 63

5 C 55

;

run;

*The created dataset looks like below -*X | Y | Z |
---|---|---|

6 | A | 60 |

6 | A | 70 |

2 | A | 100 |

2 | B | 10 |

3 | B | 67 |

2 | C | 81 |

3 | C | 63 |

5 | C | 55 |

**Example 1 : To check the distribution of a categorical variable (Character)**

Suppose you want to see the frequency distribution of variable 'y'.

proc freq data = example1;The

tables y;

run;

**TABLES statements**tells SAS to return n-way frequency and crosstabulation tables and computes the statistics for these tables.

Output : PROC FREQ |

It answers a question 'which category holds the maximum number of cases'. In this case, the category 'C' contains maximum number of values.

**Tip :**

Categorical variables are of two types -NominalandOrdinal. A nominal variable is a categorical variable in which categories do not have any order. For example, gender, city etc. An ordinal categorical variable has categories that can be ordered in a meaningful way. For example, rank, status (high/medium/low) etc.

**Example 2 : To remove unwanted statistics in the table**

**cumulative frequency and cumulative percent**to be displayed in the table. The option

**NOCUM**tells SAS to not to return cumulative scores.

proc freq data = example1;

tables y /nocum;

run;

NOCUM Option |

If you want only frequency,

**not percent distribution and cumulative statistics**.

proc freq data = example1;

tables y /nopercent nocum;

run;

NOPERCENT and NOCUM option |

**Example 3 : Cross Tabulation ( 2*2 Table)**

**distribution of variable 'y' by variable 'x'**.

proc freq data = example1;

tables y * x;

run;

Proc Freq Output |

**Example 4 : Show Table in List Form**

Suppose you do not want output to be shown in tabular form. Instead, you want final analysis to be displayed in list form (See the image below)

PROC FREQ List Form |

proc freq data = example1;The forward slash followed by

tables y * x /list;

run;

**LIST**keyword produces the list styled table.

**Example 5 : Hide Unwanted Statistics in Cross Tabulation**

proc freq data = example1;

tables y * x / norow nocol nopercent;

run;

The

**NOROW**option hides row percentage in cross tabulation. Similarly,

**NOCOL**option suppresses column percentage.

NOROW and NOCOL Options |

**Example 6 : Request Multiple Tables**

Suppose you want to generate multiple crosstabs. To accomplish it, you can run the command below-

proc freq data = example1;

tables y * (x z) / norow nocol nopercent;

run;

Thetables y*(x z)statement is equivalent totables y*x y*zstatement. In this case, it returns two tables - y by x and y by z.

**Example -**tables (a b)*(c d); is equivalent to tables a*c b*c a*d b*d;

**Example 7 : Number of Distinct Values**

The

**NLEVELS**option is used to count number of unique values in a variable.

proc freq data = example1nlevels;

tables y;

run;

In this case, it returns 3 for variable Y.

**Example 8 : Use WEIGHT Statement**

**WEIGHT statement**is used when we already have the counts. It makes PROC FREQ use count data to produce frequency and crosstabulation tables.

Data example2;

input pre $ post $ count;

cards;

Yes Yes 30

Yes No 10

No Yes 40

No No 20

;

run;

proc freq data=example2;

tables pre*post;

weight count;run;

PROC FREQ Weight Statement |

**Example 9 : Store result in a SAS dataset**

Suppose you wish to save the result in a SAS dataset instead of printing it in result window.

proc freq data = example1Thenoprint;

tables y *x /out = temp;

run;

**OUT option**is used to store result in a data file.

**NOPRINT option**prevents SAS to print it in results window.

**Example 10 : Run Chi-Square Analysis**

proc freq data = example1 noprint;

tables y * x/chisq;

output All out=temp_chi chisq;

run;

**Example 11 : Generate Bar Chart and Dot Plot**

The bar chart can be generated with PROC FREQ. To produce a bar chart for variable 'y', the

**plots=freqplot (type=bar)**option is added. By default, it shows frequency in graph. In order to show percent, you need to add

**scale=percent**. The

**ODS graphics ON**statement tells SAS to produce graphs. Later we turn it off.

Ods graphics on;Proc freq data=example1 order=freq;

Tables y/ plots=freqplot(type=bar scale=percent);

Run;

Ods graphics off;

Bar Chart with PROC FREQ |

Similarly, we can produce dot plot by adding

**type=dot.**See the implementation below-

Ods graphics on;

Proc freq data=example1 order=freq;

Tables y/ plots=freqplot (type=dot);

Run;

Ods graphics off;

**Example 12 : Include Missing Values in Calculation**

By default, PROC FREQ does not consider missing values while calculating percent and cumulative percent. The number of missing values are shown separately (below the table). Refer the image below.

Proc freq data=sashelp.heart;

Tables deathcause;

Run;

Exclude Missing : Proc FREQ |

**MISSING**option, it includes missing value as a separate category and all the respective statistics are generated based on it.

Proc freq data=sashelp.heart;

Tables deathcause /missing;

Run;

Include Missing : PROC FREQ |

**Example 13 : Ordering / Sorting**

In PROC FREQ, the categories of a character variable are ordered

**alphabetically by default**. For numeric variables, the categories are ordered from smallest to largest value.

To sort categories on descending order by frequency (from largest to smallest count), add

**ORDER=FREQ**option

Proc freq data=sashelp.heartIt is generally advisable to show distribution of a nominal variable after sorting categories by frequency. For ordinal variable, it should be shown based on level of categories.order = FREQ;

Tables deathcause / missing;

Run;

To order categories based on a particular FORMAT, you can use

**order = FORMATTED**option.

**Conclusion**

PROC FREQ is a simple but powerful SAS procedure. This tutorial was designed for beginners who have no background of any programming language. Hope the above examples help to understand the procedure crystal clear.

Good explaination!

ReplyDeleteThank you!

DeleteGreat tutorial...

ReplyDeletecould you also explain more on 'WEIGHT statement'

Good work, Keep it up :)

ReplyDeleteGlad you liked it. Cheers!

DeleteGood question bank and explanations.Keep posting!!

DeleteSimply amazing...

ReplyDeletesuperb description

ReplyDeleteGood Work, luck to stumble upon your site.

ReplyDeleteAmazing tutorial. Keep up the good work!

ReplyDeleteAmazing description. You are really making it easy for SAS beginners :)

ReplyDeleteThank you for your effort and resources

how to use by statement in proc freq

ReplyDeleteBest...I learned alot

ReplyDeletethanks sir

ReplyDeleteGood presentation sir

ReplyDeleteHi Deepanshu,

ReplyDeleteWell explained. I have 1 query regarding ods file. We are using VM space and when we are trying to save the following error occured. Could you please help.

ods rtf file = "C:\Users\kiran\table1a.rtf" style = forNESUG;

WARNING: Style FORNESUG not found; Rtf style will be used instead.

ERROR: Insufficient authorization to access /opt/sasinside/SASConfig/Lev1/SASApp/C:\Users\kiran\table1a.rtf.

Excellent explanations. Really easy to understand SAS programming through your portal.

ReplyDeleteExcellent explanations. Really easy to understand SAS programming through your portal.

ReplyDeleteThanks

ReplyDeleteVery good!!

ReplyDeleteExcellent explanations. You make me the life easier.

ReplyDeleteThis comment has been removed by the author.

ReplyDeleteThat is a very nice and well explained summary, thank you!

ReplyDelete