Choosing the Right Chart: Being Successful With Data Visualization

When it comes to visualizing data, there is no shortage of charts and graphs to choose from. From traditional graphs to innovative hand-coded visualizations, there is a continuum of visualizations ready to translate data from numbers into meaning using shapes, colors, and other visual cues. However, each visualization type is intended to show different types of data in specific ways to best represent its insight. Let’s look at five of the most common visualization types to help you choose the right chart for your data.

  1. The Bar Chart

A traditional favorite, the bar chart is one of the most common ways to visualize data. It is best suited for numerical data that can be divided into distinct categories to compare information and reveal trends at a glance.

There are a few ways to spice up a bar chart. Bars can be oriented on the vertical or horizontal axis, which can be helpful for spotting trends. Additional layers of information can be added using clustered bars or by stacking related data. Trend lines and other annotations can be added to highlight important data points. Finally, multiple bar charts could be set on a dashboard to help viewers quickly compare information without navigating several charts.

  1. The Line Chart

Like the bar chart, the line chart is another of the most frequently used chart types. These charts connect individual numeric data points to visualize a sequence of values. As such, they are most commonly used when an element of time is present. In fact, the best use case for line charts involve the display of trends over a period of time.

When two or more lines are present, line charts can be transformed into area charts by filling the space under each respective line to extend the analysis and illuminate the relative contribution that a line contributes to the whole.

  1. Pie and Donut Charts

We all love to hate the pie chart, and its cousin, the donut chart. This hatred for “dessert charts” is based on opinion and not actual research, though there are certainly many good reasons not to use these charts. Nevertheless, both are great options to visualize parts of a whole. Unfortunately, they are also among the most misused and overused chart types.

In both of these, the circle represents the 100% whole, and the size of each wedge (the largest of which should start on the upper right) represents a percentage. The trick to properly reading pie or donut charts is to not rely on the angle, but to look at area or arc-length. To avoid a bad pie chart, focus on comparing a few values (less than five, two if possible) and use distinct color separation for maximum readability.

  1. The Scatter Plot

Scatter plots are an effective way to compare two different measures by visualizing datapoints to quickly identify patterns, trends, concentrations (clusters), and outliers. These charts can give viewers a sense of where to focus discovery efforts further and are best used to investigate relationships between variables.

Adding a trend line to a scatter plot can be a helpful guide for the eye and help to better define correlation. Additionally, incorporating filters can reduce noise and assist viewers in limiting investigation to the factors that matter most to their analysis.

  1. The Bubble Chart

The bubble chart is a variation of the scatter plot that replaces datapoints with bubbles. This method shows relational values without regard to axes and is used to display three dimensions of data: two through the bubble’s location and another through size.

These charts allow for the comparison of entities in terms of their relative positions with respect to each numeric axis and size. The sizes of the bubbles provide details about the data, and colors can be used as an additional encoding cue to answer many questions about the data at once. As a technique for adding richness to bubble charts, consider overlaying them over a map to put geographic data quickly in context.


Subscribe to Big Data Quarterly E-Edition