dr2014 Tracker

Re: How do I plot line graphs for multiple groups with gplot?

dr2014 — Mon, 28 Sep 2020 17:33:58 GMT

Thank you for the recommendation.

How do I plot line graphs for multiple groups with gplot?

dr2014 — Mon, 28 Sep 2020 15:19:10 GMT

Hi all,

I need advice on how to "plot lines for multiple groups within a sas variable". I used proc lifetest to get output dataset and used proc gplot to produce an rtf file.

proc lifetest data= abc method=KM outsurv=xyz plot=(s); time duration*censor (1,2);
strata group ; run;

The variable "group" has "6 categories" and proc gplot gives me plot points by default so I used symbol statement with i=spline option, but it gave me a line graph for only one group

Here is the code:

proc gplot data = xyz;
symbol1 i=spline;
plot survival*duration= group / vaxis=axis1 haxis=axis2 legend=legend1;;
run;

How can I get line graphs for all the categories within the group? Thanks.

Adding title to gplot

dr2014 — Sat, 06 Jun 2020 10:21:42 GMT

Hi all,

I ran a proc lifetest and then used gplot to get the graph. I was able to add the axes labels and legend, however the code below does not give a title. Could you please let me know what's wrong with the code? Thanks.

ods rtf file='path.rtf';

title 'abcd';
title2 '(xyz)';
goption reset=all;*ftext=simplex;
symbol1 line=1 color=blue i=stepj;
symbol2 line=2 color=red i=stepj;
axis1 label=(angle=90 'Survival Probability');
axis2 label=('axis label name ');

legend1 label = none value = (height = 1.5) position = (top right inside) down = 2;

proc gplot data = test;
plot survival*duration = grade / vaxis=axis1 haxis=axis2 legend=legend1;;
run;
quit;

ods rtf close;

Re: Remove exponents from the values

dr2014 — Mon, 27 Apr 2020 18:08:32 GMT

Thanks for your reply @ballardw. I got the generated data step and will fix the length of the variables.

Remove exponents from the values

dr2014 — Mon, 27 Apr 2020 16:47:04 GMT

Hi all,

I imported a text tab delimited file into SAS 9.4 and the all the numeric fields are converted into an exponent of 10. So for e.g how do I convert 5.8432957E13 into the actual number 58432957310473 in the sas file now? The value with the exponent E is a numeric field with a best12. format and informat.

Thanks in advance for your solutions.

Re: Running frequency

dr2014 — Wed, 21 Nov 2018 15:44:11 GMT

Hi @KurtBremser I am back on this string. I do need help with the complete solution. Right now, this code works for me that is edited from your first code.

proc sql;

create table want as

select characteristic, count(distinct id) as n

from have

group by characteristic;

quit;

However, I need the denominator to be the 'total number of distinct ids' to get the percent. How can I accomplish that as a nested select program in sql?

so basically I need :

proc sql;

create table want2 as

select count (distinct id) as n2

from have /* original dataset*/

quit;

so this count of distinct ids are not grouped by characteristic and my percent needs to be

n/n2*100

Hope this makes sense.

I have my code as this but I know there is something missing in it

proc sql;

create table want as

select a.characteristic, a.n, b.n2

from (select characteristic, count(distinct id) as n from have group by ccs2)

(select count(distinct id) as n2 from have) b;

quit;

So I want the three columns in the final dataset characteristic, n, n2, percent

and just to expand the data:

id	characteristic
id1	RDD
id1	TSS
id2	TSS
id2	RDD
id2	GGG
id2	HHH

So the output dataset will be

Characteristic	N	N2	%
RDD	2	2	100
TSS	2	2	100
GGG	1	2	50
HHH	1	2	50

Hope this makes it all clear. Look forward to your reply.

Re: Running frequency

dr2014 — Wed, 21 Nov 2018 14:46:54 GMT

Hi @RW9 Its obvious by solution of proc freq was not helpful:

It gave an error the characteristic variable is not sorted properly.

And, the denominator for the percent has to be the total number of distinct ids.

Re: Running frequency

dr2014 — Wed, 21 Nov 2018 14:36:28 GMT

Hi @Kurt_Bremser I was looking to summarize the count but since there was a condition that could be repeated for more than an id I was concerned. Also in your code how can I ensure the counts of the condition(N) is among all ids. I mean I don't see id mentioned in the code. or that is not needed.

proc sql;
create table want as select
characteristic,
count(*) as N
from have
group by characteristic;
quit;

Running frequency

dr2014 — Wed, 21 Nov 2018 14:09:50 GMT

Hi all,

I have this dataset. I need help urgently on this.

id	characteristic
id1	RDD
id1	TSS
id2	RDD
id2	TSS

I need the final result table as:

Characteristic	N	%
RDD	2	100
TSS	2	100

how can I accomplish this

I tried by sorting the data-set by characteristic and it doesn't work.

proc freq data=have
by characteristic;
tables id/out=want
run;

Thanks.

Re: creating unique window periods

dr2014 — Wed, 21 Nov 2018 14:02:21 GMT

Thanks for your reply @novinosrinand apologies for the delay. I had to stop working on this concern as data was insufficient and had to look into more urgent tasks. Nonetheless I will test it on available data and reply.

Re: creating unique window periods

dr2014 — Fri, 16 Nov 2018 17:48:40 GMT

Please excuse the example. I hope its still understandable.

creating unique window periods

dr2014 — Fri, 16 Nov 2018 17:26:23 GMT

Hi ,

below is a table 'have'. I created the 'duration' column in it that looks at the duration between the 'enddate' of the previous observation and 'startdate' of the next record using the 'lag' function. I flagged any duration <32 as Y. For consecutive observation where the duration < 32 , thos have to be grouped together to create a unique period i.e. the dates of obs 1 and 2 have to be combined to get the new_startdate and new_enddate i.e 02/30/2017 to 8/1/2017.

It is possible that more than three obs with duration <32 could be combined to get the new_startdate and new_enddate.

Similarly for observation 3 and 4 I had the new_startdate and new_enddate i.e 10/30/2017 to 4/15/2018. I have thousands of such ids for each of which I need to do the steps to get the dataset 'want'

How can I accomplish the dataset 'want'? Please let me know. I need advice urgently.

have:

obs	ptid	index date	start	end	duration	flag
1	1	2/18/2018	02/30/17	5/30/2017
2	1	2/18/2018	6/1/2017	8/1/2017	2.00	y
3	1	2/18/2018	10/30/2017	11/30/2017	90.00	n
4	1	2/18/2018	12/1/2017	4/15/2018	1.00	y

I need to have this dataset 'want' below with the new_start and new_end columns:

obs	ptid	index date	start	end	duration	flag	New_start	New_end
1	1	2/18/2018	02/30/17	5/30/2017			02/30/2017	8/1/2017
2	1	2/18/2018	6/1/2017	8/1/2017	2.00	y	02/30/17	8/1/2017
3	1	2/18/2018	10/30/2017	11/30/2017	90.00	n	10/30/2017	4/15/2018
4	1	2/18/2018	12/1/2017	4/15/2018	1.00	y	10/30/2017	4/15/2018

Re: convert year date value from character to numeric

dr2014 — Mon, 12 Nov 2018 17:04:56 GMT

And will remember that.

Re: convert year date value from character to numeric

dr2014 — Mon, 12 Nov 2018 17:04:25 GMT

Thank you!

convert year date value from character to numeric

dr2014 — Mon, 12 Nov 2018 16:53:13 GMT

Hi,

I have the date1 value = 1978 i.e. all in 4 digits with informat and format of $64. I want to covert it to a numeric value with a format of year4.

First I tried:

data want;

set have;

if date2=input(date1,$64);

format date2 year4.;

run;

It doesn't work.

If I remove the format statement it keeps the value as character as the original variable.

I also tried date2=input(date1, year4.); and that doesn't work too.

Thanks.

Re: proc sql using where or case and when clause

dr2014 — Tue, 05 Jun 2018 02:38:27 GMT

@novinosrinThanks for your solution. It worked and was very helpful.

proc sql using where or case and when clause

dr2014 — Wed, 23 May 2018 16:47:27 GMT

Hi all,

I need some help in summarizing data using proc sql and/or any other way I could accomplish this. Below is the dataset 'have'

id	var1	var2	attempts
001	0	1	1
001	1	1	2
001	1	1	3
002	1	1	1
002	1	1	2
002	0	1	3
003	1	1	1
004	1	1	1
005	1	1	1
006	1	1	1
007	1	0	1

I need to have counts of 'var2' based on values in 'var1 and var2' and also total count of 'attempts' by each id.

So basically I need the dataset 'want' as I copy pasted below:

id	var2_novar1	var2_var1	novar2_var1	tot_attempts
001	1	2	0	3
002	1	2	0	3
003	0	1	0	1
004	0	1	0	1
005	0	1	0	1
006	0	1	0	1
007	0	0	1	1

var2_novar1: counts where var2=1 and var1=0 by unique id

var2_var1: counts where var2=1 and var1=1 by unique id

novar2_var1: counts where var2=0 and var1=1 by unique id

tot_attempts: total count of attempts by unique id i.e. if there are 3 attempts for an id then tot_attempts=3

I used the below code and was able to get the variables one at a time i.evar2_novar1, var2_var1, novar2_var1 respectively and tot_attempts.

proc sql;

create table want as

select distinct id, count(var2) as var2_novar1, count(attempts)

from have

where var1=0 and var2=1

group by id;

quit;

Is there a way I can get all variables by writing 1 sas program? I don't want to use macros to get 3 different datasets and then merge.

I tried using the code below to create 1 variable but it didn't work. If it worked I was looking to use it in getting the rest of the variables in the same program. It refers to a syntax error in the 'case & when clause' and I am unable to figure it out.

proc sql;

create table want as

select distinct id, count(case var2 when (var1=0 and var2=1)) as var2_novar1,

count(attempts)

from have

group by id;

quit;

Could anyone please help me with this? Thanks.

Re: Accessing big data using SAS PC

dr2014 — Mon, 30 Apr 2018 15:22:20 GMT

@Kurt_Bremser I will check use your code to check on the time. Thanks for all your advice. Basically, more than 1 person has contributed on the solution. Thanks much!

Re: Accessing big data using SAS PC

dr2014 — Mon, 30 Apr 2018 15:18:30 GMT

@Tom Thanks for all your suggestions. I will definitely be implementing them. I have about 20000 ids so it will be hard to put them in a macro. It would have certainly helped to subset to those ids in the first step itself.

Re: Accessing big data using SAS PC

dr2014 — Sun, 29 Apr 2018 21:58:50 GMT

@Tom I decided to use a macro to see how it works. Referring to the different librefs below is my code.I am not sure if it fits in memory, but it would be worth trying.

%macro allfiles(varlist=);

libname2.orig1&varlist

libname2.orig2&varlist

libname3.orig3&varlist

libname4.orig4&varlist

libname5.orig5&varlist

libname5.orig6&varlist

%mend;

data out_all;

set %allfiles(varlist=%str((keep=id var1 var2 type);

run;