To expand a little on Tom's explanation the PUT statement is used to write data from a SAS datastep into an object that is not a SAS dataset: usually an external file, or the SAS log or output areas.  It has the ability to write text strings and formatted values of datastep variables; with the ability to tab to a position or skip to a new line or new page. The PUT() function is used to populate a variable within a SAS dataset or view with a value derived from an existing variable or expression so it needs much less functionality.  It is always used with a format specified as the second parameter, and always results in a character value whether the existing variable is numeric or character (though SAS may have to do a type conversion behind the scenes if you specify a numeric destination). To derive a numeric value from an existing character value you need the INPUT() function with an appropriate informat.  To derive a numeric value from a numeric value you would need to nest a PUT() function inside an INPUT() function, or find a specific numeric function for the transformation you require. There are also the PUTC() and PUTN() functions which are strictly for character and numeric expressions respectively, which you need to use with %SYSFUNC() in a macro expression. They can also be used in datastep code in the case where the format needs to be selected dynamically (eg other logic in the datastep determines which format to apply). Richard temporarily back in Oz ... View more

In general and within SAS itself either single or double quotes can be used interchangeably, provided that they are used in matching pairs.  Windows users may be more used to double quotes while users of other OS' or databases may prefer single quotes.  Your choice. If quotes are located within quotes then either the internal quotes must be doubled up      'this is a dog''s breakfast'  (2 single quotes between g and s) or alternating quotes can be used      "this is a dog's breakfast". The main exception to interchangeability within SAS is that macro expressions (not just macro variables) inside single quotations are not evaluated while those in double quotes are, as has been covered by Tom.  Note that double quotes inside single quotes do not allow macro expressions to be evaluated. Where SAS has to interact with external software, eg the operating system or a relational database then the rules of the eternal application must be followed; double quotes for Windows and single quotes in most other cases.  This can be inconvenient when the user wants to pass values via a macro variable.  The answer is to 'mask' the single quote so that it is hidden from SAS but passed to the application.  One way of masking a single quote is to 'escape' it (SAS unhelpfully calls this macro quoting) thus:   %STR(%') The complete expression may need to be "unquoted" (in SAS terms) for the character to be recognised externally.  So what the OP may have been trying to achieve is possibly      ... %UNQUOTE(%STR(%')select datetime from &thisshouldwork where datasetname=&thisshouldalsowork and ... %STR(%')) ; You can define a macro function that will manage the masking:      %Macro Wrap (contents) ;           %* No semicolon on the next line! ;           %UNQUOTE(%STR(%')&contents%STR(%'))      %Mend ; So the SQL above would become      %Wrap (select datetime from &thisshouldwork where datasetname=&thisshouldalsowork and ... ) ; Richard temporarily back in OZ ... View more

This question has already been discussed in the last few days.  Please view. Richard in NZ ... View more

Use the cats() function Richard in NZ ... View more

@apple Yes you can list individual values as you suggest. I have to ask why you are storing numeric values as character variables when it is causing you some difficulty managing the data.  If your concern is saving space you would do better to apply the compress= option on the dataset.  Numeric ranges are much easier to manage intuitively with numeric formats, and this is true even if the numbers are categorical values.  Especially with dates (such as in the years data you posted elsewhere), you should be using numeric variables in preference to characters because it facilitates date arithmetic (comparing dates or displacing them by given intervals). Richard in NZ ... View more

If Proc SQL supported the median statistic I would have suggested that.  The easiest method to code would be a couple of Proc Summary (alter ego of Proc Means) statements followed by a merge: Proc Summary data = A ;      Var X ;      Output out = A_Stat           mean = Mean_XA           median = Median_XA           ; Run ; Proc Summary data = B;      Var X ;      Output out = B_Stat           mean = Mean_XB           median = Median_XB           ; Run ; Data Want ;      Merge A_Stat B_Stat ; Run ; [Not tested] Richard in NZ ... View more

Also, change the name of the column with your data in the data step Data have ;      Format Name     $200.                Date YYMMDD10.                Col1 comma12.2                ;      Set want ;      Date_Str = Substr (_Name_, 1, 11) ;      = Translate (Date_Str, '-', '_') ;      date = input (DateStr, YYMMDD10.) ;      Drop _NAME_ Date_str ;      Rename Col1 = Price ; Run ; Richard in OZ ... View more

Here are a few steps you can take Delete rows 1 to 4 and column A of your Excel table and save the worksheet Import the table into SAS - use Proc Import if you have installed the SAS Access to PC File formats Otherwise export your dataset as a CSV or Tab delimited dataset and import it into SAS (Proc Export in its Bas cut down form) In Proc Import you need to specify a table name - temp will do Apart from the Name column, SAS will give unusual names to the weekly dates, like _2004_1_5 Now use Proc transpose to change rows to columns Proc transpose                data = temp                out = have                ;      By     Noitsorted Name ; Run ; Nearly there, just need to manage the dates, now held in the _NAME_ column : Data have ;      Set want ;      Date_Str = Substr (_Name_, 1, 11) ;      = Translate (Date_Str, '-', '_') ;      date = input (DateStr, YYMMDD10.) ;      Drop _NAME_ Date_str ; Run ; Then sort it by name and date ; Proc Sort                data = have ;      By     Name Date ; Run ;      [NB untested code] Richard in NZ ... View more

apple In the case of Proc SQL it goes like this ... Select ...    stuff label = "New label for stuff" ,    ... Richard in NZ ... View more

%if %then %else are macro statements only valid within a macro (unlike say %let which can be in open code). if then else are data step statements, only valid within a data step. You have to be very careful not to mix the two, especially if you have a data step within a macro. Richard Message was edited by: Richard Carson ... View more