 |
|
| Producing Charts to Summarize Variables |
Charting Frequencies with the CHART Procedure
| Types of Frequency Charts |
By default, PROC CHART creates a frequency chart in which each bar, section, or block in the chart represents a range of values. By default, PROC CHART selects ranges based on the values of the chart variable. At the center of each range is a midpoint. A midpoint does not always correspond to an actual value of the chart variable. The size of each bar, block, or section represents the number of observations that fall in that range.
PROC CHART makes several different types of charts:
- vertical and horizontal bar charts
-
display the magnitude of data with the length or height of bars.
- block charts
-
display the relative magnitude of data with blocks of varying size.
- pie charts
-
display data as wedge-shaped sections of a circle that represent the relative contribution of each section to the whole circle.
- star charts
-
display data as bars that radiate from a center point, like spokes in a wheel.
| Creating Vertical Bar Charts |
Understanding Vertical Bar Charts
A vertical bar chart emphasizes individual ranges. The horizontal, or midpoint, axis shows the values of the variable divided into ranges. By default, the vertical axis shows the frequency of values for a given range. The differences in bar heights enable you to quickly determine which ranges contain many observations and which contain few observations.
The VBAR statement in a PROC CHART step produces vertical bar charts. If you use the VBAR statement without any options, then PROC CHART automatically does the following:
For continuous numeric data, PROC CHART determines the number of bars and the midpoint for each bar from the minimum and maximum value of the chart variable. For character variables or discrete numeric variables, PROC CHART creates a bar for each value of the chart variable. However, you can change how PROC CHART determines the axes by using options.
Note: If the number of characters per line (LINESIZE=) is not sufficient
to display vertical bars, then PROC CHART automatically produces a horizontal
bar chart. ![[cautionend]](../../../../common/64368/HTML/default/images/cautend.gif)
The Program
The following program uses the VBAR statement to create a vertical bar chart of frequencies for the numeric variable ExamGrade1:
options pagesize=40 linesize=80 pageno=1 nodate; proc chart data=grades; vbar ExamGrade1; title 'Grades for First Chemistry Exam'; run;
The following output shows the bar chart:
Using a Vertical Bar Chart to Show Frequencies
Grades for First Chemistry Exam 1
Frequency
14 + *****
| *****
13 + *****
| *****
12 + *****
| *****
11 + ***** *****
| ***** *****
10 + ***** ***** *****
| ***** ***** *****
9 + ***** ***** *****
| ***** ***** *****
8 + ***** ***** *****
| ***** ***** *****
7 + ***** ***** *****
| ***** ***** *****
6 + ***** ***** ***** *****
| ***** ***** ***** *****
5 + ***** ***** ***** ***** *****
| ***** ***** ***** ***** *****
4 + ***** ***** ***** ***** *****
| ***** ***** ***** ***** *****
3 + ***** ***** ***** ***** ***** *****
| ***** ***** ***** ***** ***** *****
2 + ***** ***** ***** ***** ***** *****
| ***** ***** ***** ***** ***** *****
1 + ***** ***** ***** ***** ***** ***** *****
| ***** ***** ***** ***** ***** ***** *****
----------------------------------------------------------------------------
40 50 60 70 80 90 100
ExamGrade1 Midpoint
The midpoint axis for the above chart ranges from 40 to 100 and is
incremented in intervals of 10. The following table shows the values and
frequency of each bar:
| Range | Midpoint | Frequency |
|---|---|---|
| 35 to 44 | 40 | 3 |
| 45 to 54 | 50 | 6 |
| 55 to 64 | 60 | 14 |
| 65 to 74 | 70 | 5 |
| 75 to 84 | 80 | 11 |
| 85 to 94 | 90 | 10 |
| 95 to 104 | 10 | 1 |
Note: Because PROC CHART selects the
size of the ranges and the location of their midpoints based on all values
of the numeric variable, the highest and lowest ranges can extend beyond the
values in the data. In this example the lowest grade is 39 while the lowest
range extends from 35 to 44. Similarly, the highest grade is 98 while the
highest range extends from 95 to 104.
| Creating a Horizontal Bar Chart |
Understanding Horizontal Bar Charts
A horizontal bar chart has essentially the same characteristics as a vertical bar chart. Both charts emphasize individual ranges. However, a horizontal bar chart rotates the bars so that the horizontal axis shows frequency and the vertical axis shows the values of the chart variable. To the right of the horizontal bars, PROC CHART displays a table of statistics that summarizes the data.
The HBAR statement in a PROC CHART step produces horizontal bar charts. By default, the table of statistics includes frequency, cumulative frequency, percentage, and cumulative percentage. You can request specific statistics so that the table contains only these statistics and the frequency.
Understanding HBAR Statistics
The default horizontal bar chart uses less space than charts of other shapes. PROC CHART takes advantage of the small size of horizontal bar charts and displays statistics to the right of the chart. The statistics include
- Frequency
- Cumulative Frequency
-
is the number of observations in all ranges up to and including a given range. The cumulative frequency for the last range is equal to the number of observations in the data set.
- Percent
- Cumulative Percent
-
is the percentage of observations in all ranges up to and including a given range. The cumulative percentage for the last range is always 100.
Various options enable you to control the statistics that appear in the table. You can select the statistics by using the following options: FREQ, CFREQ, PERCENT, and CPERCENT. To suppress the table of statistics, use the NOSTAT option.
The Programs
The following program uses the HBAR statement to create a horizontal bar chart of the frequency for the variable ExamGrade1:
options pagesize=40 linesize=80 pageno=1 nodate; proc chart data=grades; hbar Examgrade1; title 'Grades for First Chemistry Exam'; run;
The following output shows the bar chart:
Using a Horizontal Bar Chart to Show Frequencies
Grades for First Chemistry Exam 1
ExamGrade1 Cum. Cum.
Midpoint Freq Freq Percent Percent
|
40 |****** 3 3 6.00 6.00
|
50 |************ 6 9 12.00 18.00
|
60 |**************************** 14 23 28.00 46.00
|
70 |********** 5 28 10.00 56.00
|
80 |********************** 11 39 22.00 78.00
|
90 |******************** 10 49 20.00 98.00
|
100 |** 1 50 2.00 100.00
|
----+---+---+---+---+---+---+
2 4 6 8 10 12 14
Frequency
The cumulative percent shows that the median grade for the exam (the
grade that 50% of observations lie above and 50% below) lies within the midpoint
of 70.
The next example produces the same horizontal bar chart as above, but the program uses the NOSTAT option to eliminate the table of statistics.
options pagesize=40 linesize=80 pageno=1 nodate; proc chart data=grades; hbar Examgrade1 / nostat; title 'Grades for First Chemistry Exam'; run;
The following output shows the bar chart:
Removing Statistics from a Horizontal Bar Chart
Grades for First Chemistry Exam 1
ExamGrade1
Midpoint
|
40 |************
|
50 |************************
|
60 |********************************************************
|
70 |********************
|
80 |********************************************
|
90 |****************************************
|
100 |****
|
----+---+---+---+---+---+---+---+---+---+---+---+---+---+
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Frequency
| Creating Block Charts |
Understanding Block Charts
A block chart displays the relative magnitude of data by using blocks of varying height. Each block in a square represents a category of data. A block chart is similar to a vertical bar chart. It uses a more sophisticated presentation of the data to emphasize the individual ranges. However, a block chart is less precise than a bar chart because the maximum height of a block is 10 lines.
The BLOCK statement in a PROC CHART step produces a block chart. You can also use the BLOCK statement to create three-dimensional frequency charts. For an example, see Creating a Three-Dimensional Chart. If you create block charts with a large number of charted values, then you might have to adjust the SAS system options LINESIZE= and PAGESIZE= so that the block chart fits on one page.
Note: If the line
size or page size is not sufficient to display all the bars, then PROC CHART
automatically produces a horizontal bar chart.
The Program
The following program uses the BLOCK statement to create a block frequency chart for the numeric variable ExamGrade1:
options linesize=120 pagesize=40 pageno=1 nodate; proc chart data=grades; block Examgrade1; title 'Grades for First Chemistry Exam'; run;
The OPTIONS statement increases the line size to 120.
The following output shows the block chart:
Using a Block Chart to Show Frequencies
Grades for First Chemistry Exam 1
Frequency of ExamGrade1
___
/_ /|3
|**| | ___
|**| | /_ /| ___
|**| | |**| | /_ /|
|**| | |**| | |**| |
___ |**| | ___ |**| | |**| |
----------------/_ /|--------|**| |---------/_ /|--------|**| |--------|**| |---------------------
/ ___ / |**| | / |**| | / |**| | / |**| | / |**| | / /
/ /_ /| / |**| | / |**| | / |**| | / |**| | / |**| | / ___ /
/ |**| | / |**| | / |**| | / |**| | / |**| | / |**| | / /_ /| /
/ |**|/ / |**|/ / |**|/ / |**|/ / |**|/ / |**|/ / |**|/ /
/ / / / / / / /
/ 3 / 6 / 142 / 5 / 11 / 10 / 1 /
/-------------/-------------/-------------/-------------/-------------/-------------/-------------/
40 50 60 70 80 90 1001
ExamGrade1 Midpoint
The chart shows the effects of using the BLOCK statement.
![[1]](../../../../common/64368/HTML/default/images/sym01.gif){kind=small}
PROC CHART uses the same midpoints for both
the bar chart and block chart. The midpoints appear beneath the chart.![[2]](../../../../common/64368/HTML/default/images/sym02.gif){kind=small}
The number of observations represented by
each block appear beneath the block. ![[3]](../../../../common/64368/HTML/default/images/sym03.gif){kind=small}
The height of a block is proportional to the
number of observations in a block.| Creating Pie Charts |
Understanding Pie Charts
A pie chart emphasizes the relative contribution of parts (a range of values) to the whole. Graphing the distribution of grades as a pie chart shows you the size of each range relative to the others just as the vertical bar chart does. However, the pie chart also enables you to visually compare the number of grades in a range to the total number of grades.
The PIE statement in a PROC CHART step produces a pie chart. PROC CHART determines the number of sections for the pie chart the same way it determines the number of bars for a vertical chart, with one exception: if any slices of the pie account for fewer than three print positions, then PROC CHART groups them into a category called "Other."
PROC CHART displays the values of the midpoints around the perimeter of the pie chart. Inside each section of the chart, PROC CHART displays the number of observations in the range and the percentage of observations that the number represents.
The SAS system options LINESIZE= and PAGESIZE= determine the size of the pie. If your printer does not print 6 lines per inch and 10 columns per inch, then the pie looks elliptical. To make a circular pie chart, you must use the LPI= option in the PROC CHART statement. For more information, see the CHART procedure in the Base SAS Procedures Guide.
The Program
The following program uses the PIE statement to create a pie chart of frequencies for the numeric variable ExamGrade1:
options pagesize=40 linesize=80 pageno=1 nodate; proc chart data=grades; pie ExamGrade1; title 'Grades for First Chemistry Exam'; run;
The following output shows the pie chart:
Using a Pie Chart to Show Frequencies
Grades for First Chemistry Exam 1
Frequency of ExamGrade1
60 *************
**** ****
*** . ***
** . **
** . ** 50
* . *
* 14 . *
** 28.00% . 6 **
* . 12.00% *
* . . *
** . . . . **
* . .. . .. . * 40
* .. . . . .. 3 *
* . . .. . .. 6.00% *
* 5 . + . . .. . .. .1.. . .*
70 * 10.00% .. . .. . ..2.00% * Other
* .. . .. *
* . .. . *
** . . . . **
* . . 10 *
* . 20.00% *
** 11 . **
* 22.00% . *
* . *
** . ** 90
** . **
*** . ***
80 **** . ****
*************
In this pie chart the Other
section represents the one grade in the range with a midpoint
of 100. The size of a section corresponds to the number of observations that
fall in its range.
|
|
Copyright © 2012 by SAS Institute Inc., Cary, NC, USA. All rights reserved.