Data visualization is simply the graphical representation of data. If you have no idea about data visualization, check this link about data visualization examples in daily life.

According to Steele & Iliinsky (2011), data visualization has two categories in general: exploration visualization and explanation visualization. Exploration visualization is typically a part of data analysis by which you find stories in your data. The famous Anscombe's Quartet illustrates the important role of visualization in data analysis. In spatial data analysis (which can be considered as a subdomain of data analysis), the well-known Snow’s map of the Cholera outbreak in 1854 created by Dr. John Snow demonstrates the power of spatial analysis in the study of epidemiology.

Explanation visualization applies when you already know stories in your data and your try to tell others. Graphics of findings included in your presentations and publications are typical explanation visualization.

Infographic is a specific type of visualization. It usually combines statistics and graphics to narrate a story and/or to persuade audiences a specific message or argument. It has gained popularity on the web as an effective way of communication. See this article about how to increase your research impact using infographic.

Before you start

              The Designer-Reader-Data Trinity below helps you think through your design at the beginning. The first step is to determine whether your visualization is exploratory or explanatory. If it is explanatory, be aware that you create visualization for others, but not yourself. So it is important to put yourself in the shoes of your readers. It is your job to know your data and to determine what information you want your readers to learn from your visualization.

Source: Steele & Iliinsky (2011)

Know your data type

                Knowing your data type is essential for choosing an appropriate way of representing your data. There are four data measurement scales: nominal, ordinal, interval and ratio. The first two are qualitative data and the latter two are quantitative data. Nominal are known as “labels” or “names”, such as states and hair colors. Ordinal data have ordered relationship among values, but the difference between each value is unknown (e.g., satisfied, neutral, and unsatisfied). In contrast, interval data are not only ordered but also indicate the exact difference between intervals and the difference between each interval is equal. For example, the difference between 90 °F and 92 °F is the same as 64 °F and 66 °F. Ratio data have all information that interval data have. In addition, ratio data have an absolute zero (e.g., height and weight). More description can be found here.

Utilize elements in visualization

                Basic elements in visualization include position, color, size, and shape. In general, the power of these elements in visualization is ordered as position > color > size > shape (“The science of data visualization”, 2018). The image below suggests where you should put information that you want to draw readers’ most attention. Though there is no concrete maximum number of colors that a designer should use, humans can only distinguish about eight colors. So the bottom line is not to use too many colors. Watch the video “The science of data visualization” (2018) (from 15:50 to 21:07) for more suggestions and caveats of using colors. As to size, humans can tell a symbol is larger or smaller than the other, but cannot precisely determine how much they differ, especially when the shape of symbols are irregular (Steele & Iliinsky, 2011; “The science of data visualization”, 2018).

Source: “The science of data visualization” (2018)

Choose the right chart

• A cheating sheet
• Suggestions for a video preferred person: the science of data visualization (24:52 to 43:30)
• A guide for a text preferred person (see “Top 12 Most Common Used Data Visualization Types” in this article)

Tableau: Desktop , Public (free), Student version (free for one year), and free training

ggplot2 by R programming

Matplotlib by Python programming

Color Brewer 2.0: A web-based tool for generating palettes of colors.