My no-code way to label any radial in Tableau

Although it may seem like a small thing, nicely designed labels can greatly improve the appearance of a chart, especially if that chart is a radial one. Arranging labels in a circle and designing them is not an easy task in Tableau, but it is doable. In this post, I will share my way of doing it.

We will discuss how to make radial labels in two cases: when we need to label segments of a given size and when we need to label points with given degrees.

The segmented graph we’ll label is a radial heatmap of average monthly temperatures in European cities in 2019. It was created using Tableau Visionary Will Sutton’s heatmap generator:

Labeled radial heatmap.

The point graph we’ll label is a radial bar chart of the scatter of average daily temperatures in European cities in 2019. You can learn more about how to create this type of graph from Tableau Visionary Kevin Fleurlage’s tutorial:

Labeled radial bar chart.

Please follow the GitHub repo if you want to download a Tableau Public notebook with the finished graphs or the underlying datasets (I obtained the initial temperature data for them from Kaggle).

To label a radial in Tableau, we will need:

• the Ladataviz.com website for generating labels;
• the open-source vector graphics editor Inkscape for designing them.

Creating labels on Ladataviz.com

Heatmap labels

We can’t create a radial heatmap on Ladataviz.com, but we can create a radial chord diagram whose labels will be automatically placed at the centers of its segments at the appropriate angles. The idea is that we’ll create labels for a chord diagram and then substitute them into the heatmap.

Find the Chord Diagram Generator within Tools > AdvViz > Chord. This type of diagram divides a circle into segments based on their percentages and shows the flows from one segment to another. The size of each segment equals the sum of its incoming and outgoing flows. The segments are plotted clockwise, starting with ‘12 Midnight’:

Ladataviz Chord Diagram Generator.

To build the diagram, we need to upload a dataset to the site. If you click the question mark above the Choose File button, a pop-up window will describe what this dataset should look like: it should have a source column, a target column, and a value column.

Here we are not interested in flows, but in the ratio between segments to allocate their labels correctly. The sizes of the Black, Blonde, Brown, and Red segments of the above chord diagram (i.e., the sums of their incoming and outgoing flows) are 29 630, 20 230, 40 290, and 9 850, respectively. To achieve an identical arrangement of the chord segments and their labels, it is enough to specify each segment as both a source and a target of a single flow and indicate its size:

Each segment is specified as both a source and a target of a single flow.

Below is a chord diagram created from this dataset. The absolute size of each segment has doubled (as it is both the source and target of a sector-sized flow in our table), but since this happened to all segments, their percentages and locations are unchanged from the original diagram:

The flows in this diagram occur exclusively within segments.

Thus, to place labels around a segmented radial graph, we only need to know how the sizes of its sectors (intended for labeling) relate to each other. In our case, the heatmap consists of 37 equal-width segments, corresponding to the number of European cities in our dataset:

The radial heatmap consists of 37 equal-sized segments.

We can specify the size of each segment simply as 1. Or as 100 — the main thing is that the value should be the same for all segments. This is what our label dataset should look like (the full version is available on GitHub):

Heatmap label dataset (view on GitHub).

Upload the label dataset to the Ladataviz Chord Generator. Set the Width and Height to 700 and the Margin to 85. Reduce the Padding between segments to 0. Now all the labels are equally spaced and fit on the canvas:

The arrangement of the chord segments reproduces the arrangement of the heatmap segments.

Download the resulting chart in SVG format.

Bar chart labels

We know that the labels appear exactly in the centers of the chord segments. But what if we want to label several points located at certain degrees? For example, to build a radial bar chart, I calculated the angle in degrees for each bar to place it correctly. This angle is all I have now to calculate the label placement:

In a radial bar chart, we don’t know the segment sizes, but we do know the bar angles.

Here, another property of the chord comes in handy: even if the segment size is 0, the label is displayed. This means that we can label not only the segment centers but also the gaps between the segments. We will ‘fill’ the distances between our points with segments of the appropriate size. Then we’ll insert a zero-size segment into each gap and create a label for it.

Here’s how it looks with our default chord diagram. The labels should appear next to four points located at angles of 73°, 198°, 288°, and 342°:

We want to create labels for four points located on a circle at given angles.

We need to arrange the chord segments so that the gaps between them are at the specified degrees. That is, create something like this:

If we take 360 as 100%, we can calculate the sizes of the segments so that the gaps between them are located at the desired degrees.

How do we achieve this? First of all, since a chord is 100%, let’s take 360 degrees as 100%; that is, the sum of our segment sizes should be 360.

Imagine that we have a 360° dial with four hands that we need to move from zero to certain degrees. We will move them by inserting chord segments in front of them.

• The first hand, Brown, should be located at an angle of 73°. To move our hands from zero to this position, we add a segment of size 73 in front of them. Now all our hands point to 73°, where the Brown label is located.
• The second hand, Black, should point to 198°. We leave the Brown hand at 73° and move all the others to 198°; that is, add a segment of size 198–73=125 in front of them. Now one of our hands points to Brown 73° and three to Black 198°.
• The third hand, Blonde, should point to 288°. We leave the Black one at 198° and add a segment of size 288–198=90 in front of the remaining two. Now one hand points to Brown 73°, one to Black 198°, and two to Blonde 288°.
• To position the fourth hand, Red, at the desired angle, we leave the Blonde hand at 288° and add a segment of size 342–288=54 in front of the remaining hand. Now one hand points to Brown 73°, one to Black 198°, one to Blonde 288°, and one to Red 342°.
• But that’s not the end! Now the sum of our segments is 73+125+90+54=342, while the full dial should be 360°. Therefore, we add another segment after the Red hand, the size of which will make up the missing difference: 360–342=18.

Now all our arrows point exactly in the right direction.

How to apply this to our chord?

The chord will consist of two types of segments:

• Radius segments with given degrees. We will set their size in the value column as 0. Such segments will not be visible on the chord diagram, but their labels will be;
• Gap segments with calculated sizes. With their help, we will place our radii at the desired angles.

The table below shows how we calculate the Gap segment sizes based on the degrees specified for the Radius segments. It essentially follows the five-step algorithm described above:

To calculate the size of the segments, subtract the previous angle from each subsequent angle.

After calculating the Gap segment sizes, we place the Radius segments between them. This is what the final table for Ladataviz.com looks like:

After each Gap segment, we add a corresponding Radius segment.

After uploading this type of dataset to the site, it is important to select None in the Sort drop-down menu in the Nodes section. After adjusting the margin and padding, we get a graph similar to the one below. Each of its segments — both the full-size Gap and the zero-size Radius— has its own label. During further processing, we should keep only the labels of the Radius segments and get rid of the rest:

All labels except those labeling the points must be removed during further processing.

Let’s apply this method to label our radial bar chart:

Unlabeled radial bar chart.

Its labels should be identical to those of the heatmap, but now we know not the sizes of the segments but the angles of the points that need to be labeled:

City labels and the angles at which they should be located.

All points are equidistant from each other. The only exception is the arc between the last (Zurich) and the first (Amsterdam) one; it is divided into two parts by the 0°/360° point. Therefore, we need to create two segments: one will be located between 0° and the first point, and the second between the last point and 360°.

This is what the label table for Ladataviz.com looks like (the angle_degrees column is optional). Each Gap segment is followed by a corresponding Radius segment; the last Gap segment brings the sum to 360:

Bar chart label dataset (view on GitHub).

Upload this label dataset to the Ladataviz Chord Generator. Set the Width and Height to 700 and the Margin to 85. Reduce the Padding between segments to 0. Finally, in the Nodes section, in the Sort drop-down menu, select None. The segments are now sorted in the same order as in our dataset, with the Gap segment followed by the Radius segment:

We will get rid of all unnecessary labels during further processing.

We only care about how our Radius labels are placed on the canvas; we will get rid of the Gap labels in the next step of graph processing.

Download the resulting chart in SVG format.

Designing labels in Inkscape

Heatmap labels

Let’s open the heatmap SVG file in Inkscape. Below, you can see what the interface looks like. The graph will be displayed on the left side of the screen, and on the right, there will be an editing panel. At the top of this panel, click the Layers and Objects tab. Inside, you will see two folders:

Inkscape automatically splits SVG files into layers and elements; it also groups elements of the same type into folders. There is just a nuance with the Ladataviz.com chord diagram: Inkscape reads the labels along with the black polygons and places them in one folder, and the purple paths in another. Check the contents of both folders: in one, the elements will alternate as text-path-text-path; in the other, there will be only path elements. Delete this last folder as it is not needed. All the purple insides of the diagram should disappear:

Open the second folder, where the text-type elements are. Click on any element; it will automatically be selected on the graph. If you delete it from the folder, it will disappear from the graph as well.

Select and delete all path elements from the folder, leaving only the text ones:

The elements in the folder are arranged from bottom to top in ascending order; that is, at the very bottom of the folder is the first label clockwise from ‘12 Midnight’ (Amsterdam).

Bar chart labels

The only difference between processing bar chart labels and heatmap labels is that when designing the former, we need to get rid of all Gap labels:

So after you remove all the path elements, select all the text Gap elements (in my case, these are all odd numbers), and delete them. The result should be identical to that of the heatmap:

Changing the font

Further processing is the same for all radials, so I will use a heatmap as an example.

Let’s change the font of the labels. Select all the text items in the folder. To do this in Windows, hold down the Shift key and select the first and last folder items. One of the tabs next to Layers and Objects should be Text and Font; open it and its Font sub-tab. In the Font family window, you’ll find all the fonts available on your computer. If you want to add a font — for example, one of the Google Fonts — install it on your computer and restart Inkscape. The font should then appear in the Font family window. I chose Bebas Neue.

In the Style window, you can choose the font style — regular, semi-bold, bold, etc. — if you installed different versions of the font on your computer. At the bottom, choose the Font size— I chose 16. The text with the selected settings should appear in the demo panel below. Click Apply and see how our labels look now:

Changing the font color and stroke

Go to the Fill and Stroke tab. There are three sub-tabs in the tab: Fill, Stroke paint, and Stroke style. If Fill is empty and says Paint is undefined, click on the filled (Flat color) square; after that, a color wheel should appear in the tab. In the RGBA field, change the eight-digit HEX code from 000000ff to 484848ff (the last two letters indicate the transparency level; so ff is 0% transparency).

Be careful; when you click the commands, the currently selected text will be processed. Before changing the color, check if all the labels are selected:

In the Stroke paint sub-tab, we can add a stroke to the text. I want to outline the names of two cities: the one with the recorded minimum average monthly temperature (Moscow) and the one with the maximum (Madrid). In the heatmap, I use the Viridis palette, so I will take the extreme colors of its spectrum: yellow #fde725 for Madrid, and purple #440154 for Moscow.

In the Layers and Objects tab, select Madrid (in my case, it is text56). Next, go to the Fill and Stroke tab and the Stroke paint subtab. Click on the filled Flat color square. In the RGBA field, insert fde725ff. A yellow outline should appear around the label ‘Madrid’ on the canvas. We can make it thinner or thicker. To do this, go to the Stroke style subtab and set the width of the outline in the Width field. I set it to 0.5px. We do the same with Moscow, outlining its label with the color 440154ff:

Export of ready-made labels

Our labels are ready to be transferred to the dashboard. To do this, we export them as a PNG image.

In Inkscape, an object is exported using the File > Export command. Once clicked, the export options will appear in the right panel.

You need to be careful here, as Inkscape will export the currently selected object and everything on it. To export our 700 x 700 canvas and its contents, deselect the labels by clicking on the canvas. If the labels remain selected, they will not be exported within the 700 x 700 boundaries but within their own xmin, ymin, xmax, ymax boundaries. You can also place a rectangle under some of the content, select this rectangle, and export it along with the content inside it.

The current selection is displayed in the right panel. You can make it larger or smaller by changing the DPI. You can also set the Background color or leave it transparent. At the bottom, specify the file path and format, and then click Export:

I also recommend saving the SVG file itself as you work using File > Save. That way, you can reopen the SVG in Inkscape later and continue working on it.

Adding labels to the Tableau dashboard

In Tableau, we can overlay the graph on an image with labels either on the dashboard or on the graph sheet.

The simplest way is to do this on the dashboard. Just select Image in the Objects section and place it as you need, then place your graph on top of it. That is, I created a 700 x 700 dashboard, and then, when importing my label image, I specified Fit Image and Center Image. Then in the Layout section, I reduced the Outer Padding of the image and of the dashboard from 4 to 0 and from 8 to 0, respectively.

In fact, my label image became the background image for the entire dashboard. On the top of it, I placed my graph in floating mode; its parameters are 516 x 516, and the left and top paddings are 92:

The second way is to add a background image directly to the chart. This method is more complicated, but it might be useful if you want to make the chart and its background change when changing the filter.

Go to the chart sheet. First, let’s increase the chart canvas size so that the labels also fit. Our heatmap’s xmin and ymin are -1; xmax and ymax are 1. Right-click on Longitude on the Columns Shelf and check Show Header. Right-click on the axis that appears, then Edit Axis > General > Custom, and set the Fixed Start to -1.4 and the Fixed End to 1.4:

Do the same with Latitude.

Next, go to Map > Background Images and select your dataset. In the window that appears, click Add Image, then Browse and find your label image. Set the image’s xmin and ymin to -1.4 and xmax and ymax to 1.4.

In the same menu, in the Options tab, select Always Show Entire Image. Click OK.

The labels should now surround the heatmap right on the chart sheet:

On the 700 x 700 dashboard, simply specify the chart parameters as 700 x 700 and set all the paddings to 0:

I just described my method for labeling self-created radials. If you want to label one of Ladataviz.com’s radial graphs, you can download the SVG with labels in addition to the CSV or Tableau Public file, process it in Inkscape in the described way, and then upload to Tableau.

Let me also recommend a couple of useful resources on the topic:

• A video tutorial on how to export PNG files from Inkscape goes into more detail about different export parameters.
• A tutorial on how to switch an image with a selection in a Tableau Dashboard will be of help in case you want to create switchable labels.

Happy viz creating!