Skewed Data and Heat Maps in Tableau

By default Tableau uses a sequential color scheme to display the full range of values (e.g. from white to green) for a measure. But a problem comes up when the data is highly skewed (e.g. when displaying the states that students come from with over 97% being from one state and the next closest value being less than 0.5% of students). The end result of this skew in the data is that the dominate category appears in one color and all the other states show up in the same color. The trick to getting around this is to re-define how the colors are scaled across the distribution. Steps on how to do this and an example of the trick in action is available after the break… (I use a map in the example, but the same principle can be applied to any situation where color shading is used to represent values.)

To begin, you will need to switch from a sequential color scheme to a divergent color scheme. In the worksheet,

Next, under advanced options, add a check next to use stepped values. This allows you to determine the number of sets (ie colors) to be used. In most cases 10-12 should be sufficient.

Next, add a check mark for the “end” value. Now you can set the “maximum” value for the color scheme. Anything above this maximum value will be part of the last set. Set it to a value slightly above the value of the second group (e.g. state). This ensures that the top group (ie state) has the last color and the other groups will make use of the remaining colors. You can now see distinctions between the categories that previously were all shown with the same shade.

Optionally, because the data is skewed, you can “recenter” the color scheme to force a certain number of categories into the “low” color group. To do this, figure out how many groups (e.g. states) you want to include in the bottom set. Place a check mark next to center and enter a value 1 above the value for the top group of that set. This will force that number of categories into the bottom color. The remaining groups will be divided into sets across the remaining colors.