The Joy of Small Multiples
Do you know the feeling when you have prepared a beautiful chart and get this small nagging feeling that something is just not right? You would like to explore to data more deeply to see if some subset of the data behaves differently than the rest?
Say, you have prepared a chart showing the prices and sizes of sales from the local housing market, as in the chart below.
You would like to dig in deeper and see if different regions, house types, old vs. new buildings or variable like that makes the chart behave differently.
If you have data scientists programming environment available, it’s not a problem. It’s easy to use, for example, R-environment’s excellent ggplot library and create a new variant of the chart for all the subsets. Or maybe you pull out the faceting function. But if you don’t have such an environment, what to do?
When you use Power BI, you can prepare a new chart for each value of the variable using filters. It’s possible but tedious to construct. Usually, you just don’t have the time to do it. So, in many cases, this actually causes, that we don’t get all the information out of our data.
Is there a better way? Indeed there is. Small multiples to the rescue.
What are the small multiples?
They go by a different name depending on the context. Some call them small multiples, but they are also known by facets, trellis or lattice.
By all means, they are not a new invention, in fact, they were popularized by Edward Tufte’s classic Visual Display of Quantitative Information in 1983.
Well, what are they then? Wikipedia has a quite good definition.
“A small multiple […] is a series of similar graphs or charts using the same scale and axes, allowing them to be easily compared. It uses multiple views to show different partitions of a dataset.”
So, in practice, it means that you draw a new chart — similar than the base chart — with a subset of data. Which subset? Well, it depends on how you want to slice and dice your data.
Why use them?
Small multiples excel when you want to compare different subsets of data. And if you stop to think about it, it’s pretty much the basics of analytics and reporting. We want to compare the sales of different units or over time. We want to compare the performance of different units, people, machines and so on.
Here’s an example. It is the chart from the beginning of the article. It is faceted by the floor of the apartment.
It’s great, isn’t it? The human eye has such an amazing talent for finding differences in images like this. It picks in a just few seconds that most of the datapoints lie between 1–3 floors. There are only few datapoints in floor 9 and something fishy is going on in floor 8; the linear regression is steeper than on other floors.
When you really think about it, it’s amazing that we don’t utilize the power of small multiples more than we do.
Simple reason for this probably is that small multiples are surprisingly difficult to create. In Excel of Power BI, you have to create all these versions of charts by hand. It just takes too much time.
Small multiples in Power BI
Using Craydec’s custom visuals, faceting data by all variable levels is easy. You just drag a variable to the ‘facet’ datarole. That’s it.
To see it in action, here’s our Regression Chart with small multiples.
Like what you’re reading?
Want to read more? Subscribe here.
Originally published at www.craydec.com on February 17, 2019.