Per slchen you would use the input function to convert text strings into numeric variables (which are what date/time variables are in SAS).  The input requires you to specify the format of the data, length.  So in your example your date value is 10 characters, mm/dd/yyyy, spaces are interpreted as 0.  So the format to read this in would be mmddyy10.  If you have only a two digit year then: mm/dd/yy, so the format is mmddyy8.  The input reads in the text string and converts it to numeric SAS date which can have a format applied to it to show the date representation - otherwise you will just see a number.  The reverse of course is put which takes a numeric and puts it into a character string. ... View more

It looks like you are running into the display versus storage issue.  You may see in your datasets: 3039.07     and     3039.07 Which look the same.  However the data is really (an example): 3039.0700000000000000001     and     3039.07 Hence a fractional difference.  I have come up against this several times and it can happen in several different ways, most notable if the base programmer uses SAS functions to create a calculated variable, and I use SQL.  Its quite annoying.  What I would suggest is that you identify what the precision needs to be for your compare values and then explicitly round to that precision.  I.e. round(variable_from_base,8.2) = round(variable_from_comp,8.2). So round them in both datasets then do a compare again.  You could also put both variables into a character field. ... View more

Sorry, what is the question?  The output from proc compare is not always easy to read and so may require some post-processing.  Also, if you are comparing data of that size you may want to consider a bit more what you want to compare, for instance can you group the data, maybe summarise, of do by group joins to identify certain fall outs.  There are other methodologies also, some of the stats procedures will give you counts, outliers.  Distinct lists for codes, defined lists.  Maybe if you have a visit schedule, you could pull that out as a tabulate etc. ... View more

Yep, without seeing the errors am guessing, but copying and pasting the %let into the statement yields this: then catx(" ",mat.customer, input(put(%sysfunc(compress(&cmr_period ,"/",A)),4.)-1,$4.), "CMR") Which doesn't look right - e.g. what is the &cmr_period. Also, you have your inputs and put's mized up, input reads char into num, put outputs num as char. Might I suggest you do it bit by bit. e.g: proc sql;      create table want as      select     input(strip(tranwrd(CMR_PERIOD,"/","")),best.) as MY_YEAR,                    case     when ... then CALCULATED MY_YEAR                                else CALCULATED MY_YEAR - 1 end as ALTERED_YEAR,                     catx(" ",MAT.CUSTOMER,strip(put(ALTERED_YEAR,best.)),"CMR") as FINAL_RESULT ... ... View more

Hi, Or: data x;      x='01jan2012'd;      y=year(x); run; ... View more