Data visualization principle
Show the data:
One of the most important data visualization principle to begin with is showing the actual underlying dataset that is used for creating the visualization. A link to refer readers or users to the underlying dataset is a best practice to ensure reproducibility as well as increasing the trustworthiness of the visualization at hand.
Example:
In order to increase reproducibility, some visualization platforms or softwares have an efficient sharing and reproduction environment. For example:
Google Colab: a customized environment from Google that is optimized for sharing data visualization created based on Python programming. Also, this environment that resembles the Jupyter notebook has power capabilities for ensuring the reproducibility.
ObservableHQ: this platform is for developing, testing, and sharing of data visualization created based on D3.js data visualization library by using JavaScript programming language.
Provoke question about data and its meaning:
The data visualization should enable users to think about the message that is trying to be conveyed easily and bring some thoughts and new ideas about the dataset that is presented by using the data visualization. As a result, it’s important to choose clear form of plots and visual analytics modules and avoid redundant and distorted forms of charts.
Use data visualization space efficiently:
A successful data visualization should be able to transfer sufficient amount of numbers in a limited space for better clarity and sake of completeness. For large datasets, it’s important to find a efficient way to aggregate and summarize the data in order to extract the most important message of the dataset at hand and communicate it with the users through data visualizations more efficiently.
Example:
For example this visualization from Tony McGovern use tree-map for visualizing the 2016 US presidential election results at the county level in Florida that shows how to use the space efficiently and effectively at the same time:
Encourage comparison by using visual components:
An effective data visualization encourages users to compare several aspects of the plots or charts and find a more in-depth idea about most important message that wanted to be communicated through data visualization.
Example:
In this visualization of water usage for different customers, the colors and labels are chosen in a way that encourage the user to compare the his/her water usage with average, and reasonable efficient water usage to see if his/her usage is high or low in comparison to other customers and standards set municipality of their city:
interactivity and statistical analysis:
Of course, interactivity of the data visualization as well as its clear message combined with and backed up by statistical calculations would develop an effective data visualization that is able to convey the take home message efficiently to the users or audiences.
Choose colormaps to reveal details of your data:
For scientific data visualizations the way that we choose a colormap or colorbar is important to reveal the details in the dataset as much as possible. Traditionally, rainbow colormap is used for visualizing the two or three-dimensional scientific datasets, but researchers found out that colormaps should be chosen based on the details of the dataset as well as how the colors and their contrast could convey the message more effectively.
Example:
In this example using a new divergence colormap shows great details about topography of landscape in Antarctica where visualization come from Francesca Samsel with data processed using E3SM:
