More on Blue & Green pills
Tableau automatically classifies your data based on whether it contains categorical data (dimensions) or quantitative data (measures). In general, dimensions create headers when placed in the view while measures create axes.
- Blue fields = discrete (create headers)
- Green fields = continuous (create axes)
- Bold fields = sorted
- Fields with no () are Discrete (often a dimension, not necessarily)
- Fields with AGG() or something else are aggregated
- ATTR() runs something like “if MIN(var) = MAX(var) then return var”, so it’s often the maximum value.
Discrete field always add headers to the view while continuous fields add axes to the view. This distinction exists so you can use continuous dimensions or discrete measures in the view.
Aggregate calculations are always measures
Notes from webex with Joe Mako, October 2011
Tableau is a SQL generator & displayer of results. Has a built-in pipeline of how it works – this is Joe’s best guess as of Oct 2011:
- Context Filter – creates a temp table in data source, can be global or local. Only use when we want to filter out a whole bunch of stuff, e.g. testing (maybe on admissions in readmission analysis?)
- Conditional or Top N filters
- All other filters go into WHERE clause
- Then aggregations are applied:
- Filters on aggregated fields are applied, this is returned to Tableau
- Table calcs are performed
- Final layer – filtering on table calcs
- Calculate reference lines
- Then Hide, and not displaying Null marks
Null marks can be Hidden using Format pane
Values in Dimensions can be Hidden by right-click
Others get Excluded
Level of Detail Shelf:
Set to Attribute if I don’t want the Dimension to do a group by (and multiply mark results). Also useful to use aggregations to speed up processing
Reference & Averages
Reference line calculated on results is different from average calculated on table. TOTAL() function goes to underlying data, so changing a reference line from AVG() to TOTAL() will make different results
Reference lines are calculated after table calculation filters are applied
Notes from Mar 2012 WebEx w/Joe Mako
Dimensions are available to Tableau for partitioning and addressing
Aggregates return fewer marks, dimensions more. Whether aggregate or dimension affects table calcs. Discrete vs. continuous doesn’t change marks.
Aggregations will not cause partitioning, so using ATTR(Dimension) can be used to make a dimension be ignored in table calcs.
#1 Key principle in table calcs is are we working with a dimension or an aggregate.
Subtotals are based on dimension pills
What pills can be used for filtering and sorting
Current as of Tableau 7.0, calculated fields that are strings made from aggregated calcs can’t be on Filter shelf. Solution is to use a # instead of a string
describing ATTR() in more detail - Added October 20, 2014
Nice post by yours truly on ATTR()