2 Replies Latest reply on Apr 27, 2017 5:26 PM by Robert Norheim

    Time series: Multiple futures, one history.  How to structure data?

    Robert Norheim

      First time poster here.  Using Tableau 10.2.

       

      I have two somewhat related issues.

       

      In the attached workbook of climate data, note the Scenario dimension.  There are four future scenarios (rcpXX), and one 'historical' scenario.  The historical data ends in 2005, which is where the future data begins.

       

      1) I want to *always* display historical data.  Thus, the filter should just be a single value selection for the four 'rcpXX' scenarios.

      2) I want to display only the historical data that matches the filter selections for the future scenario data.  So, with the filter settings in the packaged workbook, there should only be 10 historical lines, matching the 10 lines (models) displayed past 2005.  One set of historical data (for any combination of model/metric/season) matches with up to four future scenarios for the same combination. Some scenarios have 10 models, some have more.

       

      I suspect I need to restructure the input data, separating the historical and future data. A brute force solution would be to have four copies of the historical data, one for each of the four scenarios. I'd like to find something more elegant, where I can use an inner join on the common dimensions (model/metric/season) to match up the historical data with the four future scenarios. The inner join winds up being a bit messy as it joins the record for each historical year to each future year, making the virtual dataset way larger than necessary. Not sure what I should do here, any suggestions welcome.  I've attached the original CSV dataset as well as the TWBX.