A further follow up to this issue.   The eagle-eyed may spot that my sparkline using averages, whilst looking better the figures in the data tip still did not tally with the line chart.   The line showed figures around 2000 for February 2016 whereas the line chart says the current average is just 334:     Once I spotted this even my mental arithmetic is good enough to know that (9 * ~ 2000) / 9 is not equal to 334!   I eventually figured out what is going on.   The source data I provided earlier wasn't quite complete, it was aggregated up to the Gym ID and date level which was all I needed to illustrate my question.  My full data set (attached) is finer grained than this, including a couple of other categories namely member type (3 levels) and gender (2 levels) so rather than there being one record per date there are actually 6 records per date.    My mistake was thinking that when I used this fine grained data at a higher level of aggregation (Gym ID / month rather than Gym ID / member type / gender / month) that the average aggregation would take this into account rolling up both the numerator and the denominator to this level, but it does not.  The denominator is always the number of non-missing values, so for 1st - 9th February there were a total of 54 observations (9 dates * 3 member types  * 2 genders) so the average was calculated as the sum of the measure divided by 54, instead of the sum divided by 9 as I had wanted.   In most situations once this is understood it's not a big problem, you'd just need to add a new Aggregated Measure to give the average of the measure _ByGroup_.  Unfortunately this is not one of those situations as you cannot use an Aggregated Measure in a sparkline.  As such the only solution I can see to my problem is to have a seperate LASR table (or maybe one artfully created LASR table) containing data for each grain that I want to present the sparkline at.   Hope this helps some of you avoid the same trap I nearly fell into.   Mark ... View more

I have added this suggestion to the SASware ballot, please feel free to vote it up everyone!   https://communities.sas.com/t5/SASware-Ballot-Ideas/Visual-Analytics-List-Tables-Sparkline-documentation/idi-p/249431 ... View more

Hi Michelle,   You are a genius (at puns too )  - thank you very much!  I had started tearing my hair out over that one.    I'm going to suggest to SAS that they mention this in the user documentation as it would have taken me ages to figure out that it was aggregating on a monthly basis, to me at least this was far from obvious.   Changing the aggregation to average certainly improves things, but the data tip is still confusing until you specify a somewhat long-winded name for the measure:     I also wonder if the basis for the aggregation depends on the range of the time axis.  Logically it would depend on whether a date, time or datetime item is used, but what if the range of my data was months or even more years.  Something to investigate.   Either way sparklines just became a bit more complex to explain to Report Designers.   Thanks again Michelle.   Mark ... View more