You can have the best machine learning models in the world, the tightest statistical testing, and the most robust possible conclusions, and it won’t do you much good if you don’t have the ability to communicate your results. There’s just one problem: humans are much worse at reasoning about abstract statistics than we are at reasoning about, say, baked goods. That’s why ‘30 percent’ just doesn’t mean that much to us as a number, but if we see that it’s about a third of a pie we immediately understand it. For these reasons you should want to know how to learn data visualization. This is important not only for showing non-technical folks what you’ve been up to, but also for your own processes.
As a data scientist, one of the first things I do when I have a new problem to solve is start making charts. Good visualizations are therefore key to both the maker and consumer of information.
Tools for Data Visualization
Data visualization is a specialized skill requiring you to know how to work with a specific set of tools. The basic visualization software any Python programmer should be familiar with are Matplotlib and Seaborn. These are powerful, standard issue libraries with which the aspiring data scientist can do rough-and-dirty graphing work or more sophisticated visualizations.
There are more powerful enterprise-grade options, too. It’s not necessary to know all of them, but having serious experience with at least one will do a lot to make you stand out in the job market. My recommendations are Tableau, Altair, and Plotly.
Data Visualization Resources
There’s no substitute for getting your hands dirty with one of the major visualization libraries. But there’s no reason you can’t do coursework to hurry the process along. Check out the following resources for how to learn data visualization.
The University of Illinois has a data science visualization course as part of their computer science masters, available through Coursera. It’s made up of 4 modules which can be completed in less than 20 hours, if you work hard.
The data visualization courses from LinkedIn Learning contains a lot of great information for beginners. What I like is that there’s a definite focus on the story being told by your data. People sometimes forget that there’s a key insight implied by a chart.
During and after consulting the above, make sure you find projects you’d like to work on. Projects aren’t just a great way to learn, they’re good for your portfolio as well. If you don’t have job-specific charting work to focus on, then by all means grab a basic dataset to get familiar with best practices. Sklearn, for example, has eight built in datasets that are popular for beginner machine learning projects. They’ll work just as well for beginner visualization projects.
With these pointers you can become a master visualizer, expertly communicating data-driven insights to your dazzled customers!