As for other ‘custom’ charts in Tableau, the right data structure is the key. Also, thinking the visualisation through beforehand helps a lot as well. Drawing, rethinking, redrawing is a good thing.
I wouldn’t be able to make my radar chart without following this tutorial by Tableau.
I used a radar chart in my Women & Movies analysis (blog, Tableau) to show relations between movies that pass and those that fail Bechdel-Wallace test. I set 8 measures I wanted to compare: budget, income, budget/income ratio, # votes, rating, is directed by a female, is written by a female, has at least one female lead. Each of those measures has a very different unit and scale; therefore, something has to be done about that:
I used this straightforward formula for feature scaling (min-max normalisation):
(x – min(x)) / (max(x) – min(x))
I created this calculation for each measure on average I wanted to normalise and then exported the new data to an Excel sheet fur further use.
Excel does radar charts natively (so if you need a radar chart and not a radar chart in Tableau, go for it). Tableau needs one extra row of the first measure to connect the dots and create a closed radar chart.
Another part of the prep is a radar background image. I used my normalised data to build a radar chart in Excel and then hid everything to get a clean picture.
Radar chart step-by-step
Step 1: Create the following calculated fields: angle, distance from centre and X and Y coordinates.
Step 2: Bring the Angle measure to Detail and select Compute using Attribute.
Step 3: Bring the Attribute dimension to Detail.
Step 4: Bring X to columns and Y to rows.
Step 5: Change the mark type to a polygon.
Step 6: Change mark type of Angle from Detail to Path.
Step 7: Copy X on columns and create dual axis with the synchronised axis. Change the mark type of X (2) to circle. Bring Attribute to Label mark.
Step 8: Get rid of the grid lines in formatting.
Step 9: Bring in the grid as a background image. Set the coordinates as needed to fit nicely.
Step 10: Help, the shape is weird, and the lines don’t meet. First of all, do you have that extra row in your data to connect the lines? Then try to change the sort on Attribute. (When nothing works, Stack Overflow is your Obi-Wan.)
Step 11: More formatting to make it great. Hide headers. Bring Category on colour and change opacity. Play around with the colours, so they’re easy to distinguish. Add a border, change the size of the circles of X (2). Add field labels as annotations.