Do you have an example of an expected value? And I wasn't sure what the member groups meant.
Jonathan Drummey is the author of the piece you mention on the Data Blick site so I am mentioning him as well since he may want to be aware of any information that might make the article more helpful.
Thanks for getting in touch. I have been mostly successful in adapting Jonathan’s calc fields and using them in my dashboard. I also modified them a bit to produce 25th and 75th percentiles. Currently I am weighting the data by member type(member type 1 weight = 0.25, member type 2 weight = 0.39; member type 3 weight = 2.58 ). If I filter by multiple data types things seem to be working fine, but if I select only a single member type in the filter I sometimes get 2 values as the median (see below)
Interestingly enough ,the median value I get from SPSS a number between those 2 values (e.g. 33 & 34 above for item 4, but 33.5 in SPSS). What I ended up doing was utilizing 2 sheets in Tableau. The 1st sheet gets used if multiple member types are selected in the filter (using the weighted median calculations.) If the user selects a single member type in the filter a 2nd sheet is used with uses the MEDIAN Tableau function. Since a single membership type is being filtered I don’t have to apply a weight in the calculation of the median. I would still like to figure how to get the weighted median calc filed to work properly for my purpose. Since I am applying the weighted median calculation to multiple items I am also trying to figure out how to use either the “mean” interpolation or the “high” interpolation depending on whether the median value falls between 2 data points. I hope this make sense.
image003.jpg 16.9 KB
I'd seen your earlier post as well and haven't had a chance to get back to this, I'll take a look over the weekend.