How can you use visual variables to encode data for exploratory data analysis?

1 What are visual variables?

Visual variables are the basic components of any data visualization. They are the attributes of the graphical elements that you use to encode your data, such as shape, size, color, position, orientation, texture, and so on. Each visual variable can convey different types of information and has different perceptual properties. For example, position is a very accurate and intuitive visual variable for encoding quantitative data, such as coordinates or values on a scale. Color, on the other hand, is a more expressive and flexible visual variable for encoding categorical or qualitative data, such as groups or labels.

2 How to choose visual variables?

The choice of visual variables depends on several factors, such as the type and structure of your data, the purpose and audience of your visualization, and the available space and tools. However, there are some general principles and guidelines that can help you make informed decisions. One of them is the principle of expressiveness, which states that a visualization should use the minimum number of visual variables to encode the maximum amount of information. Another one is the principle of effectiveness, which states that a visualization should use the most appropriate visual variables for the task and the data. For example, if you want to compare the values of two variables across different categories, you might use a bar chart with position and color as visual variables, rather than a pie chart with angle and color.

3 How to evaluate visual variables?

One way to evaluate the effectiveness and expressiveness of visual variables is to use Bertin's framework. Bertin was a French cartographer and graphic designer who proposed a classification of visual variables based on their perceptual properties and their suitability for different types of data. He identified six main visual variables: position, size, value, texture, color, and orientation. He also defined four levels of data: nominal, ordinal, interval, and ratio. He then mapped the visual variables to the data levels according to their ability to show identity, order, quantity, and difference. For example, position can show all four levels of data, while color can only show identity and order.

4 How to apply visual variables?

To apply visual variables to your data visualization, you must first identify the type and structure of your data. Is it quantitative or qualitative? Is it univariate, bivariate, or multivariate? Is it discrete or continuous? After that, define the goal and question of your exploratory data analysis. What are you trying to discover or communicate? What are the main variables or dimensions of interest? Then select the appropriate chart type and layout for your data and goal. For instance, a histogram or boxplot could be used to explore the distribution of a single variable, while a scatterplot or line chart could be used to explore two variables. Afterwards, choose the visual variables that best encode your data and support your goal. For example, position, size, and value could be used for quantitative variables, while color, shape, and texture could be used for qualitative variables. Finally, adjust the visual variables to enhance readability and aesthetics. Contrast, saturation, hue, labels, legends, and axes can all be utilized to provide context and scale.

5 How to learn more about visual variables?

If you’re interested in learning more about visual variables and how to use them for exploratory data analysis, you should consider checking out The Grammar of Graphics by Leland Wilkinson. This book provides a comprehensive and systematic framework for creating and understanding data visualizations based on visual variables and algebraic rules. Additionally, Jacques Bertin's article Visual Variables summarizes the classification of visual variables and their properties and applications. To further your knowledge, you may benefit from Data Visualization: A Practical Introduction by Kieran Healy, which instructs readers on how to use R and ggplot2 to create and customize data visualizations using visual variables and layers. Lastly, Data Visualization: Charts, Maps, and Interactive Graphics by Robert Grant covers the theory and practice of data visualization using visual variables and examples from various domains and tools.

6 Here’s what else to consider

This is a space to share examples, stories, or insights that don’t fit into any of the previous sections. What else would you like to add?