I have a demographic data set with no missing data that covers a 12 month period. It is built like the below:  data a; input date :monyy. age sex race count; format date monyy.; datalines; JAN18 24 0 0 1 JAN18 22 1 1 3 FEB18 19 1 2 1 MAR18 20 0 0 4 MAR18 20 1 4 2 ; RUN; The categorical variables are numerical with values representing different levels (e.g., sex=0  is male, sex=1 is female; race=0 is white, =1 is black, =2 is asian, etc.). Count means how many individuals with that combination of preceding variables are in that month sample. There are no individual identifiers in the sample.   The total number of individuals and distribution of levels of variables changes slightly each month, and I'm wondering how best to find the one month average for each variable and value due to some other monthly analyses.   In other words, I want to average the 12 month period demographics to one month for a "table 1" since I don't have individual information of how they may come and go over the study period (i.e., if there are 12000 people in the entire study, the one-month average n=1000; if there are 9000 white people over the 12 month period then white people would represent 750 of the one month sample, etc.). I would want this to round to the nearest whole person.    Any ideas on how best to code? I appreciate the help.  ... View more

Hi. I have a SAS dataset (A) with dates formatted into mmddyy10. format. I have another data set (B) originating from a CSV with a 4 digit numeric variable confirmed to represent YYMM (i.e., 20YYMM).  If merging these two to eventually do an analysis by year and month, how would you recommend formatting the data set B numeric variable?     data A; input event date :yymmdd10.; format date yymmdd10.; datalines; 1 2020-01-01 2 2018-07-06 3 2015-02-15 ; RUN; data B; input var yearmonth; datalines; 1 1801 2 1911 3 2202 ; RUN;         ... View more

I have a large dataset with several variables. Where record=C, this is the current accepted Label value. Where record=H these are historical labels.   I am having trouble figuring out how to produce a new dataset with the label for every ID that is current (i.e., where record=C), but also if there was a change in label during the time period. Obviously getting where records=C is easy, but that 'if there record value changed during the set time period' part is tripping me up.    In other words, if there were historical changes but the label didn't change, then only the current record is desired. If there was a label change I would want both the current and the historical record to see what the label was changed from .    Time period of interest 2018-01-01 through 2023-12-31  (*note, present record dtend=9999-12-31)   data have; input record $ label $ dtstart :yymmdd10. dtend :yymmdd10. id; format dtstart yymmdd10. dtend yymmdd10.; datalines; H WA123 2010-01-01 2018-07-05 1 C WA123 2018-07-06 9999-12-31 1 H AJ456 2015-02-15 2019-10-30 2 H AJ456 2019-10-31 2022-12-31 2 C AJ456 2023-01-01 9999-12-31 2 H AJ123 2000-01-25 2010-08-20 3 H AJ123 2010-08-21 2020-09-01 3 C AJ124 2020-09-02 9999-12-31 3 ; RUN; data want; input record $ label $ dtstart :yymmdd10. dtend :yymmdd10. id changeflag; format dtstart yymmdd10. dtend yymmdd10.; datalines; C WA123 2018-07-06 9999-12-31 1 0 C AJ456 2023-01-01 9999-12-31 2 0 H AJ123 2010-08-21 2020-09-01 3 1 C AJ124 2020-09-02 9999-12-31 3 1 ; RUN; Does my want make sense?  For IDs 1 and 2 : I don't need previous records because label didn't change. For ID 3 I need the current and the previous record (which I would want flagged as being part of a changed set) because they changed during the time period.   The last caveat, if there were multiple label changes in the time period for an ID I would need to see each of them, not just the most recent.  ... View more

I have a data set with two dates. The first indicates when an event happened (date1). The second when an event resolved (date2). Resolution can only take place on a weekday due to business hours.    I'm trying to figure out the best way to code in a data step to measure days until resolution (timetores) if: 1) I decide I want to not count the first weekend days only if date1 was on a weekend (i.e. weekday in (1 7)).   In other words, if date1 was on a Saturday or Sunday and date 2 was on a Monday I would want timetores to =0, as Monday was the first day with the opportunity to resolve. This timetores should be the same as if date1 and date 2 were on the same weekday. In either scenario, days after the first Monday or subsequent weekday would add +1 day.    Weekend days following the first weekend, or any weekdays if Date1 occurred on a weekday, would add +1 day as with any other day.    2) Alternatively, how would I code if I decide I want to not count any weekend days, regardless of when Date 1 took place.    In this example I basically wouldn't penalize days where resolution could not take place. So if Date1 took place on Friday, the next Monday should start with 1. If date1 took place on a Monday, the following Monday timetores would =5 (Tue=1, Wed=2, Thu=3, Fri=4, Mon=5).    Does this make sense?        ... View more