Hi ballardw, Many thanks for your reply. With one variable I mean one column. It is one big character/string column that holds the codes. Example: --id--bigcolumn-- __21__7051-3984,7514-7594,6843-6848,6677-6685,1212-6561 "if so it may well be worth the not-too-complex effort to restructure the data to to a single code value" but we never know how long the code string will be. It could be a 100 columns or just 10. We would need to loop through each string and create columns dynamically. @TomKari, Thanks for your advice! "In one case, you use a four digit number as the item of interest. In the other one, you use the two numbers separated by dashes" Sometimes they are seperated by dashes, sometimes by comma. Whats is sure is that each is 4 digits. "Is the data always in the form "9999-9999,9999-9999...", or can it vary?" -> It can vary. I am not sure how many rows we receive by day but its around 10m per year. Right now we dont have that big column with codes in it. We have a much simpler structure and it is stored in one SAS dataset. Good question about the reporting. We want to analyse data according to the individual codes in the big code string. For example, give me all data with "chicken+packed+meat". I can build dimension tables with those codes and their text value. No problem but how do I efficiently look through the facts that have their dimension-keys all in a single column.
... View more
