Data Visualization: Creation process

 

1. Define the purpose, audience, and context

First, you need to identify why you are creating a visualization in the first place, and what message you are trying to convey to your specific audience, given your contexts. 

Questions to ask yourself:

• What is the purpose of the visualization? Is it exploratory (end goal to analyze data to detect trends, outliers, clusters) or explanatory (with key takeaways)? 
• If explanatory, is the purpose to advocate and persuade the audience or to convey a scientific or factual message?
• What is the main takeaway for the audience?
• Who is the audience of the visualization (age group, demographic, country)?
• Where will the data visualization be shared (published article, online, poster, presentation, etc.)?
• What is the context for creating the data visualization, in terms of timeline, level of familiarity with tools, and budget and available resources?

 

2. Collect, explore, and clean data

Cleaning data is often forgotten about in the data process, but it's often said that 70% of working with data is cleaning the data, 30% is analyzing and visualizing it. There are several ways to clean your data, from basic Excel, to OpenRefine, to programmatically doing it in R or Python. 

Questions to ask yourself:

• ​​​​​What kind of data do you have? Is it microdata, data at an observation/individual level, that you need to aggregate to visualize? Is it data in a csv tabular form or is it geospatial data with shape files? 
• Where is your data from? Do you have multiple sources of data that you need to join on a common variable? (Ex. GDP per country and life expectancy per country, joined on the country name)
• Do you need to clean your data? Are there missing data, outliers, or mismatched formats? (Ex. dates are inconsistently formatted, you have both "null" and "N/A" to represent missing data).

 

3. Select visualization type and tool

Based on your purpose, audience, context, and data, what type of data visualization is best? This subject guide introduces some basic types of data visualization. Other tools to learn about data visualization types and decide which one best suits include:

• The Data Visualization Catalogue and its Chart Selection Guide
• From Data to Viz decision tree 
• The Python Graph Gallery 
• Directory of Data Visualizations by Claus Wilke
• Interactive Chart Chooser by Depict Data Studio

The type of visualization type will help narrow down a tool choice. Tools should also be chosen based on where and how the visualization will be displayed and your level of familiarity with tools and coding. This subject guide introduces tools depending on your data type and needs.

 

4. Test, refine, and iterate visualization

The first visualization you create or that a tool creates for you may not be its ultimate form - there are many things you can think about and change when taking into consideration:

• Elements of your visualization like the title, subtitles, description, axes, or legend
• A different type of visualization or a different tool to better meet your needs
• Colour schemes, accessibility, and readability
• Design principles like the Gestalt principles
• Changing your variables and data

 

5. Share and publish visualization 

The most exciting step - sharing your data visualization with the world! Depending on your context and place of publication, sharing your visualization before publishing it can also be very important as a feedback tool to continue to refine your work. 

Things to consider before publishing:

• How to cite the data you have used
• Incorporating feedback so the message you are trying to convey is understood by viewers