topic Re: proc sgplot for datasets with holes in Statistical Procedures

proc sgplot for datasets with holes

Babado — Wed, 22 Sep 2021 13:05:53 GMT

I have this dataset frequencies

bucket	freq	date
1	0.49%	202004
1	0.43%	202005
1	0.48%	202102
1	0.46%	202103
2	1.99%	202004
2	2.31%	202005
2	1.89%	202006
2	2.70%	202007
2	2.17%	202008
2	1.72%	202009
2	2.33%	202010
2	2.27%	202011
2	2.60%	202012
2	3.00%	202101
2	2.37%	202102
2	3.06%	202103

genearted by

data frequencies;
    input bucket $1. freq NLPCT7.2 date $6.;
    format freq NLPCT7.2;
    datalines;
1	0.49%	202004
1	0.43%	202005
1	0.48%	202102
1	0.46%	202103
2	1.99%	202004
2	2.31%	202005
2	1.89%	202006
2	2.70%	202007
2	2.17%	202008
2	1.72%	202009
2	2.33%	202010
2	2.27%	202011
2	2.60%	202012
2	3.00%	202101
2	2.37%	202102
2	3.06%	202103
;
run;

The time window should go from 202004 (April 2020) to 202103 (March 2021) but there are some missing frequencies for bucket 1.

When I run the sas program

proc sgplot data = frequencies;
	title "frequency by date";
  	series x=date y=freq / GROUP=bucket;
	xaxis fitpolicy=rotatethin;
run;

I get the following plot

which, although I know the problem comes from the fact of missing data, I don't understand.

Is there a way to overcome this?
I would like to have both series plotted for their respective time window if possible. If not possible, I would like the missing frequency values to be assumed 0%.

Re: proc sgplot for datasets with holes

PaigeMiller — Wed, 22 Sep 2021 12:35:36 GMT

I'm not getting the plot you are getting, so do us all a favor (this time, and at all times in the future). Please provide the data as SAS data step code, following these instructions: https://blogs.sas.com/content/sastraining/2016/03/11/jedi-sas-tricks-data-to-data-step-macro/ Do not provide data as Excel files or screen captures, provide data only as SAS data step code.

Re: proc sgplot for datasets with holes

Babado — Wed, 22 Sep 2021 13:07:35 GMT

Thank you for the advice, as this is my first time using this platform. I edited the post.

Re: proc sgplot for datasets with holes

Ksharp — Wed, 22 Sep 2021 13:16:34 GMT

Store your date as numeric/date type variable ,NOT character.

data frequencies;
    input bucket : $ freq : NLPCT7.2 date : yymmn6. ;
    format freq NLPCT7.2 date yymmn6.;
    datalines;
1	0.49%	202004
1	0.43%	202005
1	0.48%	202102
1	0.46%	202103
2	1.99%	202004
2	2.31%	202005
2	1.89%	202006
2	2.70%	202007
2	2.17%	202008
2	1.72%	202009
2	2.33%	202010
2	2.27%	202011
2	2.60%	202012
2	3.00%	202101
2	2.37%	202102
2	3.06%	202103
;
run;

proc sgplot data = frequencies;
	title "frequency by date";
  	series x=date y=freq / GROUP=bucket;
	xaxis fitpolicy=rotatethin;
run;