We are living in a world where data are ubiquitous but getting data insight is hard. Data visualization provides quick and ready to understand information for large data sets. Since data visualization has great power and with great power comes great responsibility. There are many ways to visualize the same data. There are various tools like Excel and other similar programs which can be used to make perfectly acceptable visualizations, or terrible ones.
The tool can help, but it’s ultimately up to the user to make decisions. So what is the difference between a good visualization and a bad visualization then? I would argue that a good visualization clearly and accurately conveys the key messages in the data. A bad visualization will obfuscate the data either through ignorance or malice. So what does this mean? Visualizations can be used by an analyst for their own consumption to gain insights into the data. Visualizations can also be used to provide information to a decision maker and/or to convince someone of something.
Now, a bad visualization can hide patterns that could give insight or mislead decision makers. This is where the malice part comes in. We will look at a few examples of visualizations taken from nationally published reports. We shall report these examples as case of bad visualization but we state its due to ignorance not malice. This ignorance is due to very little attention paid to data visualization as these visualizations are by experts who are trained in data analysis and models but know little about this visualization aspect of data.
My focus is what is good and what is bad about them. Then we will build better versions of these for ourselves. Nevertheless, visualization is inherently subjective and the right visualization will depend on the situation. So use your own judgment and think about what I talked about before with a good visualization and a bad visualization.
Our focus here is on the abuse of pie chart and use of 3-D. It is often used but since area comparison is always difficult so better to avoid it for more than 3 to 4 categories. Following are examples of abuse of pie charts. These charts are from Quarterly Infrastructure Finance Review of SBP, Pakistan.
This pie chart is constructed for a report by the SBP on Infrastructure and investment. This chart show percentage of disbursement in various sectors during Apr-Jun, 2016. In visualization, less is more. When we consider this chart, we must focus on key given on the left column and then we move right to match the color with the number. This makes underlying information difficult to grasp particularly if you are giving a talk to your management who has to make decisions based on this information. Moreover, it becomes difficult to differentiate between close colors like blue, orange and light orange. Additionally, this graph does not indicate values properly for purple and light blue between green and blue at the bottom. It requires good effort to grasp the information while sole objective of data visualization is to get quick picture of the data. Another example of bad visualization is given as follows from the same report.
Since I don’t have access to the exact data and I have used this pie chart data as per my best effort and have constructed following bar chart (it depends on audience which visualization is easy to understand).
Another mistake which we make while using pie chart or bar chart is use of 3D chart.
And of course, the 3D-effect on the pie chart adds nothing, but does play a subtle trick on the eye. Due to the 3D-effect, the blue and red segments are larger looking, which at a glance, may lead the viewer to overestimate their size. I have made a simple bar chart for this data.
Another aspect of charts is to minimize non-data ink ratio. Chart given below shows complete data as well as bar charts. If table is used instead of bar chart, it will provide all information making either table or chart redundant.
Moreover, purpose of this chart is probably to show performance of the sectors over time which may better be reflected with the following bar chart.
This post will help to avoid excessive use of pie chart and for more stuff on pie chart read “Save the Pies for Dessert” Stephen Few, Perceptual Edge Visual Business Intelligence Newsletter August 2007
Note: Purpose of the blog is to create awareness about the correct use of data visualization since dataviz is a mean to make sense of data.
• The Analytic Edge, http://www.edx.org
• Quarterly Infrastructure Finance Review http://www.sbp.org.pk/departments/ihfd/QR/Infrastructure/2016/Apr-Jun-2016.pdf