Data Visualization

Key Takeaways from OpenVisConf 2017

Bookended by presentations from two of the “titans” of open-source data visualization, Mike Bostock (creator of D3.js) and Hadley Wickham (RStudio chief scientist and keeper of the “tidyverse”), Bocoup’s fifth OpenVis Conf was tightly packed with talks that both informed and inspired.

Before introducing his new, reactive code environment,, Mike Bostock shared a passage from Bret Victor’s Explorable Explanations, which nicely frames many of the challenges and opportunities addressed at the conference, and in data/information visualization in general:

“An active reader asks questions, considers alternatives, questions assumptions, and even questions the trustworthiness of the author. An active reader tries to generalize specific examples, and devise specific examples for generalities. An active reader doesn’t passively sponge up information, but uses the author’s argument as a springboard for critical thought and deep understanding.”

One can replace the word “reader” with viewer, or user, but the sentiment remains the same— visualization, as a medium for communication, inherently exists in a broader context. As Amanda Cox (New York Times) described in her keynote on day two, there is always a gap between “what you say and what people think you say.”

This is especially true when it comes to probability and uncertainty. Of course, this is not a problem restricted to visualizations. Amanda included a chart from a study of “Words of Estimative Probability” conducted by CIA analyst Sherman Kent among NATO officers in the 1950s, which illustrates just how subjective and varied individual interpretations can be.

Source: VisualCapitalist <>


The choices one makes in visualization will impact how information is received. Therefore, there is no one “right” software or framework— the answer is almost always “it depends.”

Catherine D’Ignazio & Rahul Bhargava discussed visualization tools as “informal learning spaces.” They offered best practices for targeting “learners” rather than just “users,” by making tools and interfaces: focused, guided, inviting, and expandable.

The importance of accounting for context and audience(s) was a theme in a number of the presentations. At times, this might mean accommodating an environment in which several different tools are used. Amy Cesal’s talk, “Why does data vis need a style guide?”, explored some of the ways in which style guides can be used not only to creates a sense of brand coherence, but to free people up to focus on the content of what they’re doing, rather than having to make decisions about colors, typographies, etc. again and again.

In a similar vein, Connor Gramazio introduced one example of a “computational design assistance tool,” colorgorical, which “automatically creates palettes based on user-defined balance of color discriminability vs. aesthetic preference.” The aim of such tools is to help build best practices into the pipeline from data collection, to analysis, interpretation, and communication.

To wrap up the conference, Hadley Wickham covered the “Role of Visualization in Exploratory Data Analysis (EDA). ” He quantified his Github commits over time and visualized them by day of week, time of day as well as whether he was traveling or not. Not surprisingly, he codes much less when he’s traveling. R coders were clued into the incredibly useful forcats package, which does everything from grouping factor variables together to shifting orders of factors on plots.  

Overall, it was an excellent conference filled with a diverse set of speakers. Everyone left feeling highly energized to put their newly acquired data visualization ideas to work.