Data Visualization: Choosing Simple and Effective Charts
Although you have many chart types to choose among when presenting data, it’s good to start with some of the simple and most commonly used charts for the most chance for success: bar and column charts, line charts, and pie charts.
Bar and column charts
Some people use the term bar chart when speaking about a chart that shows the data horizontally or vertically; others call a chart that displays the data vertically a column chart. Whatever you call them, these charts are best used for comparisons.
The figure below shows an example of a column chart. Notice that the chart is simple, with a title, a labeled axis, and clear labels to show what the columns represent.
When you use a column chart, be sure to shorten or use smaller labels on your x-axis below each bar to ensure they display horizontally. Utilizing longer labels will result in needing to display the title vertically (as shown in the figure), which is hard for the user to read.
A line chart connects data points over a period of time, as shown in the following figure. Line charts are best used for something like a trend to show movement. These charts are easy to read and fairly easy to create. This type of chart should be one of your staples.
The use of pie charts is controversial, and the debate is more than a decade old. Just type the words avoid pie chart in a search engine, and you’ll literally find more than 1 million entries. One of the best-known data design experts, Edward Tufte, refers to pie charts as “dumb” in his book The Visual Display of Quantitative Information (Graphics Press). Tufte argues that pie charts are dumb because they fail to show comparisons and trends as well as bar or line charts do. Many experts argue that the eyes are not good at estimating areas, which you must do when viewing a pie chart.
However, you can use pie charts as effective data visualizations if — and only if — you stick to the purpose they were meant to serve and follow the guidelines in this section.
By definition, pie charts are circular charts divided into slices, with the size of each slice showing the relative value. In other words, at a glance, it should be easy to see which slices of the pie contribute the most and least to the whole pie. Well, it’s not quite as easy as you may think.
Take a look at the two most common ways pie charts are misused:
Too many slices are displayed. You should limit the number of pie slices to five. Displaying additional slices that are too small to be sorted will only distract the user from the main point. The following figure shows a pie chart displaying how much (by percentage) each revenue stream has contributed to the company’s overall revenue in the last quarter.
At a glance, it’s clear that T-shirts, capris, and baseball caps combined account for 90 percent of the company’s sales. What isn’t so clear are the products that make up the remaining 10 percent of the revenue.
The figure below shows a better way to display the same data. Notice that the other products are combined in a slice titled Other. This makes the chart easier to digest. You highlight the top contributors and show the contributions of the additional slices as a single sector.
Slices of equal value are displayed. This is another common mistake. The pie chart in the figure above has fewer than five slices, but because the value of some of the slices are relatively the same, it’s hard to compare the actual contribution of those individual slices relative to another one another.
The following figure displays the same data from the preceding figure in a column chart that has been set to sort in ascending order.
Notice how much easier it is to see which products have contributed the most revenue, even if the differences in some of the values are very slim?
Unless you’re developing static data visualizations such as infographics or a yearly report in which the data isn’t updated dynamically, avoid using pie charts. The reality is that most Big Data visualizations are going to be updated dynamically from some real-time database, making it nearly impossible to control the data output. The risk of breaking one, if not both, of the rules of pie charts provided in this section is very high; ultimately, the risk isn’t worth making the data hard to read.