There’s a saying that birds of the same feather flock together. When you throw your data on a scatter chart, data with similar characteristics stick together and some data points land far away from the bunch. Data points that appear together represents the norm in your data set, and those that fall at a distance on the scale allows you to identify outliers, mavericks, the very angry birds. Outliers can either be good or bad that require further analysis.

In this post, let’s learn how to create meaningful scatter charts in Power BI Desktop.

The Data

To create a scatter plot, you need at least two measures and one dimension in your data set. The following table – only a portion is shown – that I use for this demonstration contains movies released in 2014. The table has two measures: TotalGross and Production Budget, and four dimensions: Movie, Year, Legend, and Studio. One dimension would have been enough, but I included additional dimensions to show examples to make the scatter chart more visually appealing.

Creating the Scatter Chart

After importing the data into your model, in the report view, click on Scatter Chart in the Visualizations pane. It’ll create an empty chart. With this empty chart selected, drag TotalGross and Production Budget to X Axis and Y Axis respectively. You should see a single circle representing the aggregated value of these two measures as shown below.

Next, drag-and-drop Movie to Details. The chart should appear as the one shown below. If you look closely, the number of circles equals the number of rows (movies) in the table.

As you can see, several circles overlap and appear in buckets meaning that they follow similar pattern for budget spent and gross collections. A few appear away from the buckets meaning that they’re different from what can be considered the norm.

Related: Create a Pareto (80-20 rule) Chart in Power BI

Beautifying the Scatter Chart

There are few options you can use to enhance the scatter chart. Note that these options doesn’t work for every situation, but I’m showing here so you’re aware of different things you can do to enhance the scatter chart.

Size the Circles

For example, drag and drop TotalGross to Size and you’ll see the size of circle change based on the measure, TotalGross; movies with big gross collections get bigger circles. This works in some cases, say finding the top grossed movie, but try to find the 10th highest grossed movie. Human eye can’t naturally interpret the sizes. There’re other charts – bar charts for example – that work well for these scenarios.

Use Legend

There is a calculated field called Legend in the table which has three possible values: Ugly, Good, and Too Good. This field is calculated based on the variance between TotalGross and Production Budget. If the movie collected less money than the budget, the Legend would be Ugly, and if the movie collected more money it’d be either Good or Too Good.

Add Legend field to Legend in Details of the visualization. In the Format pane (brush icon), expand Data Colors and select appropriate colors for these three categories. I select Red, Blue, and Green for Ugly, Good, and Too Good respectively. As you can see in the below image, the circles are now color coded helping you to focus based on a criteria.

The above image which uses variance as a Legend looks pretty, but it highlights what’s already obvious. For example, we already know that any circles in the lower left of the chart did poorly at the box office. Color coding them in Red doesn’t provide any additional insights. So, another option you can consider is to format based on Studio. In the following image, movies released by Warner Bros (WB) are in yellow and all other movies are in blue. This nicely compares movies released from WB with the rest.

Final Thoughts

Scatter charts are awesome. They’re best used to find correlation between two measures. As you saw in this example, scatter charts also help you identify outliers in your data, and with all the different options available you can create stunning scatter charts using Power BI Desktop.

Also see using scatter charts to create a calendar report from Dustin Ryan.