I do not understand what you mean--can you provide more detail, and maybe some screenshots or a sample workbook?
Well I have a table
currency | monday | sunday | tuesday
GBP | 80 | 12 | 30
EUR | 25 | 8 | 5
If I create a pill SUM(Sales) and put it on Marks shelf as Colour it will create gradient scale and the numbers will be coloured. However I want 80 and 25 to be at the same level because they are the best in their respective currency groups. That's what I mean by group application of colour scale. Each group (whether it's column or row) should have its own scale
I'm not sure what your data looks like exactly, but may be able to use a WINDOW_MAX(Sum(Sales)), placed on the Color Shelf, with Compute Using set to "Currency". If you set up a stepped, divergent color scheme, it may work for your scenario, but I don't know what your production workbook or your data source looks like. Here's an example done with sample data:
That makes sense. I have seen Compute before just didn't put two and two together. However my colour pill doesn't have Compute Using option which seems odd..
My data looks exactly like yours I just have currency as product, and day of the week as date. so basically I would like to set up a colour scale and each currency will have 7 values (1 dark green, 1 green etc...)
Each row will effectively contain 7 values of 7 different colours
I had to write a table calculation to do this. It was just WINDOW_MAX(Sum(Sales)), placed that on the color shelf, set the compute using, and you can set up 7 stepped colors, believe. I'll try to create an example, I was using the Beta when I created this so I couldn't upload the workbook.
Here's an example. There are probably other ways to do this, some may even have better options, this is just what popped in my head when I saw the thread. I'm not an expert on this topic, just trying to learn new things myself.
I would strongly suggest studying Jonathan Drummey's guide to Conditional Formatting in Tableau here:
Color Scale Window Max ML.twbx.zip 159.7 KB
It worked, I guess with very different data (sales from 1 to 100000$) the scale will be a bit skewed.
I've got 7 values in each row. Which is fine and let's say the values are:
the biggest four are obviously practically the same colour. How can I use stepped scale to make sure all seven values are different enough so the best one is recognised quickly?