Visualizing Data
Data tells a story.
Once you discover the story, the challenge becomes communicating it to others. Data visualization is a great way to present information in a way that your audience can easily understand.
Tables and graphs are two primary ways of presenting data. Both are great for communicating, but in different ways.
Tables have columns and rows filled with text and numbers that are read vertically (columns) and horizontally (rows). Tables display precise values and make it easy to look up detailed information. While tables are great for displaying exact values, they're not very good at highlighting trends or relationships between those values.
Graphs communicate visual patterns and relationships through lines, bars or other objects. There are many different types of graphs, and each uses the physical attributes of length, position, size, shape and color to communicate their message. The use of these visual aids makes graphs very effective at communicating trends between values.
There are many different types of graphs and choosing the right one for your data can be a challenge. Remember to always use the type of chart that communicates the context of your data and makes sense to your audience. Here are a few options to keep in mind:
Line graphs plot a series of data points and connect them with straight line segments. Line graphs are great for showing how one continuous variable trends across equal intervals of time. Line graphs are very adaptable and multiple lines can also be used to display several reporting a of the same variable across time and in the same scale.
Horizontal bar or vertical column graphs excel at communicating differences between non-continuous, categorical data. Bar and column graphs display values as rectangular bars with lengths proportional to their values. They should always start at zero and are great for showing relationships between observations of one variable, or different observations across related variables. More complex comparisons are possible by grouping or stacking bar charts.
A pie chart is a circular graph divided into segments like a sliced pie. Pie charts communicate proportion through the size of each slice and are best used when there are three or four things to compare, and they sum to 100 or 100%. The problem with pie charts is that it's difficult to accurately judge the size of circle fragments versus the rectangles of a bar or column graph. Because of this issue, pie charts should be used sparingly, if at all.
Donut graphs are essentially pie charts with the center cut out. This makes them even more difficult to read than pie charts. Think twice before using them.
Scatter plots are a more complicated type of graph that visualizes a positive, negative or lack of correlation between two continuous variables. Each point in a scatter plot represents a single observation. Color can be used to group observations and enhance communication. Trend lines can can also help show the relationship among the observations.
Maps can be used when your data is geographical, although a bar or column graph may show relationships better.
Special effects available in most charting programs should be avoided, as they rarely add value to your visualization. 3D in particular skews the relative size and shape of your data, making it difficult to understand the relationship between observations. Avoid using 3D effects and minimize your use of special effects in general.
Data visualization is only successful if your audience understands the story of the data. Tell the story, provide context for the data and deliver actionable information to help your audience be successful.
For more detailed reading, I recommend Stephen Few's book, "Show Me the Numbers," or "Data Points" by Nathan Yau.
