5 Ways to Improve Data Visualization

Linniar Tan
4 min readSep 23, 2018

I have been researching on data storytelling concepts and best practices to effectively show our findings to the audience. There are many compelling and eye-opening practices that I found, so I decided to write a summary that I can always refer back to every time I have to create my own data stories.

This summary is mostly based on an amazing book from Cole Nussbaumer Knaflic, Storytelling with Data: A Data Visualization Guide for Business Professionals, and the work of Edward Tufte in his book, The Visual Display of Quantitative Information.

1. Use Color Sparingly and Strategically

Colors can be a powerful way to draw our audience’s attention to the things that we want them to focus on. In her book, Cole mentioned about the power of pre-attentive attributes, like color, size, and position to signal what’s important.

Region with the most apartments is highlighted in blue

2. Adding Details through Annotations

We can use annotation to describe relevant external factors. Using the same color also makes it easier for the audience to see which part of the data is relevant to the description.

Annotations are highlighted with the same color as the bar to make it easier for audience to see which part of the data is relevant to which annotation.

3. It’s Recommended to Use Line Charts for Continuous Data

Line charts are usually best for continuous data, while bar charts work great for categorical data.

It might be tempting to use bar chart
But, line chart gives us a cleaner view of how the trend looks like

4. Maximizing Data-Ink Ratio (Declutter the Graph)

In his book, Tufte says: “The larger the share of a graphic’s ink devoted to data, the better”. Ink that fails to depict statistical information does not have much interest to the audience and therefore should be erased. It’s surprising that there are a lot of graphics with redundant and non-data-ink that can be erased without affecting the information that we want to convey to the audience. By erasing these “junks”, we can even convey the information in a much better and cleaner way.

Source: http://www.storytellingwithdata.com/

A few questions that we may ask: Do we need those gridlines? Do we really need the texts to be in bold? Do we need that y-axis if we can put on the data labels? If we have findings that we want the audience to focus on, should we go ahead and highlight it with color, while we turn other categories into gray?

I was mesmerized when I saw the transformation from an unattractive, busy charts into a much more minimalistic, effective charts. Definitely check out these links to see how they are amazingly “salvaging” these charts.




5. Things to avoid

Never use 3D charts! Unless we are actually plotting a third dimension. 3D charts are not as cool as we thought and they tend to lead us to misinterpretation of the data because they are more difficult to see.

The 3D effect on the left bar chart leads to a misleading information.

We also need to think twice when we are tempted to use pie charts. There are always better alternatives to pie charts. Pie charts do look nice, but they are not as effective as other charts.

In pie chart, it is difficult to see how each category compares to each other, especially when the percentages are more or less the same. Bar chart definitely does a better job!

This summary has been a useful reminder for me every time I need to create my own data visualization. I will keep updating it every time I find cool concepts and practices that are worth noting!