What You'll Learn About Data Visualization in a Data Science Master's Program

Data visualization is the art and science of presenting abstract statistical information in an understandable and informative visual format to make it easily accessible to viewers.

Data scientists can uncover incredible and useful information from the large datasets they often work with, but presenting that information in a way that normal users can easily interpret and understand has always been a challenge. The insights are often worthless if they can’t be conveyed to a board of directors or anyone else responsible for making the final decision on how to put the findings to work in the real world.

Data visualization, or just data viz, as it is often called in data science circles, is the solution to that problem. Although most data scientists will need to exercise data visualization skills in their career, not all of them are good at it. Good data visualization, like programming and analysis, requires creativity, but it also requires skill in visual design and a degree of artistic talent.

If a picture can be worth a thousand words, a well-crafted data visualization can represents millions of database records.

Visualization is About Making Information Available to Those Who Need It

Humans are visual creatures. David Hyerle, a pioneer in mind-mapping research, estimates that 90 percent of the sensory input to the brain is visual information. And the human brain is extraordinarily well-adapted to processing that information: around 30 percent of the brain’s cortex is devoted to vision, versus 8 percent for touch and 2 percent for hearing. Around 60 percent of all brain function is often involved in the complete chain of vision from perception to recognition to reaction.

Text, of course, is a visual representation of data. But it is a complex one, patterns of patterns, that require additional processing to decode into meaning. Shapes, colors, and simple lines, on the other hand, are easier to process. If similar meaning can be imparted through either format, the purely visual will always be faster and easier to comprehend.

Turning data into such pictures can take many forms, some of the most popular being:

  • Bar charts
  • Scatter plots
  • Streamgraphs
  • Heat maps
  • Tree maps

But good data visualization is limited only by the skill and imagination of the person creating it. Unlike most data science professions, it doesn’t necessarily require an in-depth and formal education to be good at data visualization. Although you have to understand what the underlying data represent, the deep technical details about how it was initially derived don’t necessarily matter.

And in the reverse of that case, it’s not always necessary to have strong artistic flair or design skills to produce solid data visualizations today, either. Online tools like Tableau or programming packages like chart.js make putting together relatively attractive, accurate data visualizations easy for even the least artistic data scientist.

Although packages like Tableau and chart.js can automate simple types of data visualization, the truly iconic and memorable data visualizations usually require some custom design and programming work.

How Data Scientists Turn Data into Pictures

The classic example of good data visualization is the original London Underground map. Created by an electrical draftsman in 1931, the map threw away elements once considered important, such as geographic layout and distance, in favor of emphasizing relative organization, important stations, and simple directionality. By reducing the complexity of the information, the map became more readable and at the same time delivered the data that consumers were most interested in. It was an instant hit.

Computers, of course, allow faster, more flexible, and more subtle presentations of data. A clean analog interface like the Tube map is also necessarily limited in the information it can present. A digital representation, however, can be fractal—clicking on points of a visualization to allow drilling down into detail, or even exploring related data, is becoming a common feature in business intelligence visualizations like in Amazon’s QuickSite visualization suite.

Today, more and more data visualizations involve animation rather than just static images. This real-time global wind force map is both hypnotic and effective, for instance.

Nor are data visualization experts stopping at simple animations when it comes to telling stories with data. This visualization of the traffic forces that causes buses to bunch up on busy routes has become interactive, taking on game-like elements that involve viewers in the process of telling the story of the data being presented.

Working With Data Visualization as a Data Scientist

The sophistication and multi-disciplinary skills required to create such complex, yet captivating, visualizations, takes them out of the hands of the average data scientist and into the realm of specialists. There are different job titles being used for such professionals in different fields, including:

  • Data Visualization Developer
  • Data Visualization Engineer
  • Infographics Engineer
  • Business Intelligence Analyst

These professionals work with tools like:

  • Plotly
  • Tableau
  • Charts.js
  • Raw
  • Visual.ly
  • Nodebox

Many data visualization experts don’t have any formal training as data scientists, but those who do have far more tools at their disposal for digging into complex data sets and pulling out representative illustrations. A formally trained data scientist with a master’s degree has probably been exposed or taught how to use R or Python to make direct connections to data sources, sift or process the data, and turn it into visual charts.

All data viz experts will find themselves working with other professionals, either those who are producing or consuming the information. Understanding the data that executives and other target audiences need to see is vital in producing a useful visualization.

For data science experts who aren’t also great artists, they might also turn to actual artists or graphic designers to help put together effective, attractive visualizations. Simple concepts such as color charts, white space, and weighting aren’t generally taught in data science programs, but they can make all the difference when it comes to putting together a visualization that is genuinely effective. Using professional designers to consult on or create the basic design can be a game-changer.

Whether you end up working in the field as a data scientist or branch out exclusively into data visualization, you can be assured that whatever data you are working with will be more useful and get more attention if you use visualization techniques to present it.