Tag Archives

3 Articles

We bring you pie: the value of 3D in data visualization

Personally I do not like the use of 3D in data visualization. Every time I see a 3D column chart I can’t help thinking that who made it must have something to hide. The data might not be interesting enough so the graph needs to spice things up.

In my opinion 3D is pretty much useless and more a distraction than a help. There is one exception to this and that is the third dimension in a XYZ plot like the one below:

However the general usefulness of these plots (however sexy) is virtually none.

So, bottom line: I see no value in 3D in data visualization.

Or, better, I saw no value in 3D in data visualization. Because now we have 3D printing! Suddenly there is value in 3D with BI. How cool would it be to actually be able to print your 3D graph and have an actual model of the graph? If you make it from degradable substances you could even re-use the substance when you refresh your data and reprint your graph.

Think about it for a second! Let’s for once assume that people will still be using pie charts (despite my rant against them). I see a business model here that finally makes the pie chart live up to its name and have some value (fill a stomach or two).

“Pie your data: your bring us data, we bring you pie”. Or “eatable insight”. What I mean is this: a company that delivers delicious pies that show your data. Let’s imagine you plot your sales amount per region as a pie chart.

You order the pie and have it delivered to your sales meeting. Each will get the slice for their region.
How’s that for direct feedback? “Feeling stuffed? That’s because you have done so well!”. “Hungry? Make sure you sell more and you will earn yourself a bigger slice of the pie!”.

I hate pie charts and so should you

Pie charts. By far the most popular chart type among business users. Also the most unusable chart type. Every time a business users asks me for a pie chart a kitten a bit of me dies. In this post I explain why.

Let’s start with the basics: what is a pie chart? A pie chart is a circular graph that is divided in parts (slices) by lines coming from the center. It is used for part-of-a-whole analysis, such as to compare product sales per product category. Here it is important to be able to spot small differences in size between slices. When multiple slices need to be shown it is often displayed in 3D (yuck!) because otherwise the smaller slices would not be visible.

To explain what is wrong with pie charts, let’s imagine a clock. A normal clock is a circle with two hands to show time, such as the one below:

Look at the clock above: it quite clearly shows 5 o’clock right? It also divides the total circle into two segments, one obviously larger than the other. All is fine here.

Let’s look at the same clock again:

Now the clock seems to show some minutes just before 4 o’clock, maybe 3:55? Here it already gets harder. That’s why we added numbers along the circles perimeter to help you read time, like this:

Ok, let’s now imagine a clock with three hands (often used to show seconds):

See how the clock divides the circle into three sections, all of almost equal size. Here, let me label them for you:

Care to tell me which one is bigger? A, B or C? Which one comes in second? Which slice is the smallest? Pretty hard right?

Let’s add another hand and bring the number of slices to four:

Now, suppose you want to understand which one is the biggest and which one is the smallest. The biggest is obviously either A or D, but which one? The smallest is clearly B or C, but again, which one? You can’t tell.

However, for a user, it is vital to know if Product A sold more than Product D and by how much. Also, if you want to understand which product is the worst-performer and why how much we need to be able to judge if B or C is the smallest and by how much.

Of course, we can come up with a work around and include data labels to help the user understand:

Now we can see that A is the best performing product while C is the worst performer. However, we had to add labels to be able to see it.

The point here is that as a species we humans are practically incapable to determining if a slice of a circle is bigger than the other. If only given a short glance some people will not see a 10% difference. If given more time most people with see differences as small as 5%. What if the difference is 1%? What is that 1% meant profit or loss for your entire company?

The examples above are very simple: I hardly come across pie charts with only four slices. Imagine what happens when there are more! What happens if one slice is really big (let’s say 90%) and there are ten others dividing the remaining 10%?

There is something seriously wrong with pie charts. Data visualizations should be built to tell the story immediately and accurately. The user should not have to give it a minute or even 20 seconds to study the graph to understand what is going on. The user should be able to spot even the smallest of differences while still seeing the big picture. Pie charts may look pretty but are impossible to read quickly and accurately. (This holds true for most pie charts. It gets better if only two slices are shown, much like a clock).

For part-of-a-whole I prefer to use bar charts and depending on the data I go for a 100% stacked bar chart, like the ones below:

 

So, please get rid of the dreaded pie charts. Tell users that they should not be looking at pies but eating them. Make this world a better place and eradicate the pie chart once and for all!

 

 

Taking the new version of Power Map for a spin

One of the most exciting tools in Excel has recently released a new version! Get the updated version of Power Map here: http://www.microsoft.com/en-us/download/details.aspx?id=38395 (Excel 2013 only).

This update comes with a ton of new features, including some of the most popular feature requests I have come across.

One of the first things that you will see when you add data to Power Map is that Power Map is now more intelligent and automatically suggests how to interpret columns. In the screenshot below it automatically mapped the ‘Gemeente’ (County / Municipalty) and ‘Provincie’ correctly. This goes to show that this automatic mapping not only works with English terms J

 

The way you choose between types of column charts has changed. You just choose the ‘Column’ type and then on ‘Category’ you can change the chart sub type (clustered or stacked).

Power Map can now also use calculated columns and supports hidden fields. Annotations can now not only contain custom text or fields but can also display an image.

A new visualization type is ‘Region’, which is actually really cool. It provides shapes as available from Bing to overall items on the map. As far as I can see it not only works for the obvious countries, but also for example for provinces in the Netherlands and even the municipalities.

If you have chosen a category for this new type of visualization you can also change the way regions are shaded, for example by relative values in the same category or across all items.

Then, with a tick of a button you can change from a 3D map to a 2D / flat map (note the cool animation).

Also, it is now possible to save your tour to a video right in the tool. Video types can be optimized to presentations and HD displays, computers and tablets and for mobile devices.

One great new feature is that it is now possible to change the coloring of charts, regions and bubbles:

Here is a sample tour I created in 20 minutes based on population data that I loaded from CBS (Central Bureau of Statistics in the Netherlands) using Power Query (hint: make your selection in CBS’ Stat line and use the “from web” option in Power Query to get the data in Excel).

 

%d bloggers like this: