Choosing your color is a significant factor in the development of powerful charts in Tableau. The story that you want your data to tell will be highlighted by a good selection of colors, while a bad one will distract the user from the actual purpose of the visualization.
In this article, we will focus on defining the types of a color palette that we can use while creating data visualization in Tableau. Also, we will include some general tips and best practices when working with color. Let’s learn together the importance of color and become proficient in choosing a color that will get the best out of our visualizations.
Three General Types of Color Palette
The palettes of color that we commonly use for data visualization can be divided into three general categories:
Remember, the type of color palette that you decide to go for in your reports highly depends on the nature of the data that should be identified with that color. That is why the first thing you need to answer is – what is the main goal of your data visualization. What are you planning to achieve with this data visualization? Once you answer this, you can easily choose the purpose of your data visualization – and as a result, the accurate color palette.
When the value, on which you are planning to base your visualization, is categorical by nature, a qualitative palette should be used. Categorical variables are those that, without natural ordering, are listed under distinct names. Nation, state, product, or gender are just a few examples. In other words, dimensions in Tableau are values that can be listed as distinct. Since all the values are listed as distinct, a color from a qualitative palette is applied to each potential value of the attribute individually.
The colors assigned to each category need to be distinct as well, in the qualitative palette. You should aim to restrict the overall palette size to ten or fewer shades, as a rule of thumb. By including more colors than suggested, you are consciously and potentially running into the problem of differentiating between groups.
This is an example where you can use a qualitative palette in Tableau when creating visualizations. We created a pie chart using a qualitative palette to show the total sales by Region. We chose this palette Region as a dimension since the members of this field are distinct and each of them can be aligned with a different color.
But even though Region is a dimension, or simply said has distinct members, does not mean by default that it is appropriate to use in this chart. If the dimension has more than 10 members, the chart will look overwhelmed and not useful insight can be obtained from it. For example, State is also a dimension and all its members are distinct, but it is not appropriate for this chart, since it has a lot of members. The chart looks chaotic, even though all members have their color.
Due to their shades, we can make a distinctiveness between colors. Also, by changing lightness and saturation we can achieve greater variation between colors, but it is best if you don’t make the differences between colors too large. Too much variation could mean that certain colors are more meaningful than others, though if this kind of practice is made on purpose, this can be a valuable asset to your visualization. If the qualities associated with certain colors are related, is ok to have colors with the same hue, but different lightness and saturation. Another way, try to avoid this practice.
As an example, if you tracking your sales per product for the last month, you can use few light-colored lines for different products, while on the other hand use a darker shade for the line that should represent the total number of products for the last month. In this case, by using a darker shade you are trying to emphasize the line since it is the total one. In meanwhile, due to the lighter shade of the other lines, the user can get a picture that those lines are part of the total number of products.
If the values that are assigned to be colored are numeric or have values that are naturally ordered, then a sequential palette will be used to represent them. Simply said, we cannot list each value as an individual one, they come as one chain of values. A chain of continuous values. In Tableau, these values are known as Continuous Measures. Sales and Profit are just two examples of fields that contain continuous values, that cannot be separated from each other. In a continuum, colors are distributed to data values, usually dependent on lightness, hue, or both.
For a sequential palette, the most popular color dimension is its lightness. Typically, lighter shades are correlated with lower values and darker shades with higher values. This is, though, because the data view on Tableau appears to be white background or a very light one. It is normal to have the opposite case on a dark background if applicable, where higher values are identified with brighter, lighter colors.
For a sequential color palette, the second important dimension is its hue. For your color chart, it is always good to only use a single hue, often varying lightness to show meaning. However, as an additional aid in encoding, spanning between two colors is something worth exploring. Usually, a warmer color would go on the lighter end (a color toward red or yellow), while on the darker end a cooler color will take place (toward green, blue, or purple).
Let’s create a visualization in Tableau where we will use a sequential palette. We want to create a simple data table where we can see the profit and sales that we generate from each sub-category. Simply data table will give us numbers, but we can not spot a winning or losing sub-category on the first look. That is why we will use the field Profit for color. So basically, since Profit is a continuous measure, its members are not distinct values and they can be described as a ‘chain of values’. That is why, for the sub-categories where we have higher profit, a darker shade of color will be applied. As a contrast, a lighter shade will be applied for the sub-categories where we don’t generate enough profit. As we can see, Copiers is our most profitable sub-category while Tables is our least profitable sub-category.
If our field with numerical values has an important central value, then we should choose a diverging palette. A diverging palette is actually the fusion of two sequential palettes with a common endpoint, residing at the central value. Colors on one side of the center are assigned to values greater than the center, and colors on the opposite side are assigned to values lower than the central one.
For each of both sequential palettes, usually, a distinctive hue is used to make it easier to differentiate between positive and negative values relative to the central value. As with sequential palettes, a light hue is normally applied to the central value, so that darker colors suggest a greater distance from the middle.
In order to explain the usage of the diverging palette, we can use the same example as above. By using a sequential palette, we pointed out at Tables as our least profitable sub-category. But even though, the profit for this sub-category is negative, in other words, there is no profit, still, it receives the lightest shade of the color used, not something unique. One way to distinguish the profitable sub-categories from the non-profitable ones is by using a diverging palette. This palette will color the cells for the non-profitable sub-categories red, and those that have profit, green. That way we can immediately spot the sub-categories which don’t bring any profit to us. Since the number that distinguishes between negative and positive profit is 0, we need to state 0 as a middle so Tableau can read the numbers properly.
Tips When Using Color In Your Visualizations in Tableau
Avoid using color on places when it is not necessary
Although the color is an integral part of the data visualization itself, it is best to maintain restraint and use color only where appropriate. Not every visualization that you make will need several colors. If you have only two variables to plot, the vertical and horizontal points or lengths are likely to encode them.
Color typically only comes in when it is required to encode a third variable into a chart or if it is a part of a specialized chart, a pie chart. There are cases, though, where color may be applied to illustrate a particular finding or as an additional encoding spotlight.
Try to be consistent with color across visualizations
It is a smart idea to align colors between charts when they apply to the same category or entity, whether you have a dashboard or report that contains several charts. If colors constantly change across visualizations, you are also losing their meaning. As a result, this will make it more difficult for the user to understand the chart or the report, in overall.
Leverage color meaningfulness
Sometimes, you can use the colors as they are commonly perceived to increase the effectiveness of the visualizations. In other words, to leverage from the common perception of a certain color. When the entities for who you are working, have underlying color norms, such as for sporting teams and political parties, it will be easier for readers to understand a visualization by assigning the suitable colors. You may also want to try to build personalized palettes around the colors of your brand as a basis.
To minimize eyestrain, a common rule of thumb is to prevent excessive usage of high levels of color saturation and brightness. By giving them a bolder look compared to the other elements, can make space for highlighting the significant elements. Similarly, in order to place unimportant details in the past, among other reasons, the value of gray should not be neglected.
It is worth mentioning, as a final thought, that various cultures will connect different meanings with each color. For instance, in some Western cultures, red may be synonymous with passion or risk, but in some Eastern cultures, wealth and good fortune. If studies are addressed to a wide audience, this may not be especially relevant, but it is another method to bear in mind to better render your visualizations.
Don’t forget color blindness
Around four percent of the population, most of them males, have some form of color blindness. While there are also cases of color blindness that cause blue and yellow colors to appear the same, the most prevalent forms of color blindness cause inconsistency between certain shades of red and green.
For these purposes, in order to show the meaning associated with a certain color, it is useful to experiment with other dimensions other than hue alone. For example, lightness and saturation. To get an understanding of whether the final visualization would be understandable to others and whether there are possible ambiguities, you can even use colorblindness simulators which can be found online.
This article offers a short description of how color can be used for powerful data visualization. Depending on the form of data mapped to color, various types of the color palette can be used – qualitative, sequential, or diverging. Make sure that the color that you’ve to choose is meaningful and used correctly. To maximize the accessibility of your visualizations, strive to work with colorblindness. When explaining results to others, always strive to carefully evaluate your color options, since a successful selection of colors would make it so much easier to convey your intended message to your audience.
By the Editorial Team
By the Editorial Team
By the Editorial Team
By the Editorial Team