topic Split into deciles based on dates in SAS Programming

Split into deciles based on dates

Jarvin99 — Fri, 22 Apr 2022 16:13:04 GMT

I want to assign deciles for tweets within each publishday. How can I do it?

Below is my code:

proc sort data=abc.twitter;
by publishday tweets;
run;

data try2;
set abc.twitter;
keep publishday tweets;
run;

proc rank data=try2 groups=10 out=try2;
var tweets;
ranks decile;
run;

The result I get is only 8 ranks instead of 10 ranks each day. I use proc freq to check out the number of ranks within each day. But only 1,3,4,5,6,7,8,9 are shown, and rank 0 and rank 2 are missing.

proc freq data=try;
table decile;
by publishday;
run;

Thank you.

Re: Split into deciles based on dates

Reeza — Fri, 22 Apr 2022 16:29:20 GMT

Check your data for duplicates.

Re: Split into deciles based on dates

Jarvin99 — Fri, 22 Apr 2022 16:48:25 GMT

No. I just checked. I have unique firm-date data for tweets for a firm on a day. Now, it becomes only 8 ranks after I rerun the same codes. Really don't know why the results are inconsistent.

Re: Split into deciles based on dates

PaigeMiller — Fri, 22 Apr 2022 16:51:50 GMT

@Jarvin99 wrote:

No. I just checked. I have unique firm-date data for tweets for a firm on a day. Now, it becomes only 8 ranks after I rerun the same codes. Really don't know why the results are inconsistent.

The duplicates you need to check for are in the number of tweets, not in the unique publishday

Another possibility why you are getting only 8 ranks is because there is only 8 data points for a unique publishday

Re: Split into deciles based on dates

Reeza — Fri, 22 Apr 2022 17:25:09 GMT

Aren't tweets text data? How are deciles being calculated for that? Or should it be based on time of day for the tweets?

You have your deciles on the variable tweets. You're saying the variable tweets doesn't have any duplicates? So no retweets in the data?