topic Re: Line graph with 95% confidence interval in SAS Programming

Line graph with 95% confidence interval

sandrube — Wed, 05 Jan 2022 20:36:11 GMT

Hello friend,

Happy New Year!I have the following data set and am trying to create a line graph with 95% CI. I used the proc sgplot (below). But it gives 95% CI as shade. Can anyone help me tell me how to do 95% CI as bar, not as shadow.

data have;
input Month_year $ frequency StdDev ColPercent ColStdErr ColLowerCL ColUpperCL positive negative;
cards;
Jan_2018 27580 729.40706 1.0132 0.0182 0.9776 1.0488 0.0356041171 0.0356041171
Feb_2018 25225 678.96555 1.0562 0.0195 1.0179 1.0944 0.0382764006 0.0382764006
Mar_2018 28230 746.38890 1.0956 0.0194 1.0576 1.1335 0.0379820518 0.0379820518
Apr_2018 26760 700.25026 1.0778 0.0188 1.0410 1.1146 0.0368349788 0.036834978
May_2018 27550 742.60845 1.0735 0.0195 1.0353 1.1118 0.0382656688 0.038265668
Jun_2018 25035 679.67669 1.0136 0.0187 0.9770 1.0502 0.0366360408 0.0366360408
Jul_2018 26160 716.09459 1.0352 0.0195 0.9970 1.0734 0.0381609915 0.0381609915
;

proc sgplot data=have;
band x=Month_year lower=ColLowerCL upper=ColUpperCL / transparency=0.5;
series x=Month_year y=ColPercent ;
run;

Thnaks

Muni

Re: Line graph with 95% confidence interval

Reeza — Wed, 05 Jan 2022 21:05:18 GMT

proc sgplot data=have;


scatter x=Month_year y=colPercent / yerrorlower=ColLowerCL yerrorupper = ColUpperCL ;
series x=Month_year y=ColPercent ;
run;

Use a SERIES plot instead of BAND. You can customize it further using options on the SERIES statement, to hide the marker or change colours or add/remove endcaps etc.

See the docs for all options:

https://documentation.sas.com/doc/en/pgmsascdc/9.4_3.5/grstatproc/p1lcbd3lhs3t3bn1jk6d8sjt2yqx.htm

@sandrube wrote:

Hello friend,

Happy New Year!I have the following data set and am trying to create a line graph with 95% CI. I used the proc sgplot (below). But it gives 95% CI as shade. Can anyone help me tell me how to do 95% CI as bar, not as shadow.

data have;
input Month_year $ frequency StdDev ColPercent ColStdErr ColLowerCL ColUpperCL positive negative;
cards;
Jan_2018 27580 729.40706 1.0132 0.0182 0.9776 1.0488 0.0356041171 0.0356041171
Feb_2018 25225 678.96555 1.0562 0.0195 1.0179 1.0944 0.0382764006 0.0382764006
Mar_2018 28230 746.38890 1.0956 0.0194 1.0576 1.1335 0.0379820518 0.0379820518
Apr_2018 26760 700.25026 1.0778 0.0188 1.0410 1.1146 0.0368349788 0.036834978
May_2018 27550 742.60845 1.0735 0.0195 1.0353 1.1118 0.0382656688 0.038265668
Jun_2018 25035 679.67669 1.0136 0.0187 0.9770 1.0502 0.0366360408 0.0366360408
Jul_2018 26160 716.09459 1.0352 0.0195 0.9970 1.0734 0.0381609915 0.0381609915
;

proc sgplot data=have;
band x=Month_year lower=ColLowerCL upper=ColUpperCL / transparency=0.5;
series x=Month_year y=ColPercent ;
run;

Thnaks

Muni

Re: Line graph with 95% confidence interval

yabwon — Wed, 05 Jan 2022 21:09:07 GMT

Do you mean something like that:

proc sgplot data=have;
HIGHLOW    x=Month_year LOW=ColLowerCL HIGH=ColUpperCL / TYPE=BAR transparency=0.5 BARWIDTH=0.5;
series x=Month_year y=ColPercent ;
run;

Re: Line graph with 95% confidence interval

Reeza — Wed, 05 Jan 2022 21:26:04 GMT

FYI - I would recommend changing that first column to a SAS date instead of a character and then having it formatted for display.