Why Visual Analytics is a solution to the data talent gap (and the data language gap)

It’s more than just having data expertise in the organization

In recent years, more organizations are addressing the “data talent gap” and are responding to it by offering training programs in analytics and data science. Assuming more people go through training as analysts and data scientists and enter the workforce, more organizations may benefit from having more deep analytical talent in the next few years.

However, I am skeptical that the data talent gap problem will be solved by that alone. Rather, I see that the “gap” will shift from the gap caused by the lack of data experts in an organization to the gap caused by the lack of communication and understanding between data experts and non-data experts in that organization. In essence, the data talent gap becomes a “data language gap.”

Typically, not everyone in the organization is an analyst, and it can be difficult for data experts to communicate the value of an analysis to non-data experts who can potentially be the domain experts or even the decision makers.

This is where Visual Analytics (VA) can help tremendously. VA is “the science of analytical reasoning facilitated by interactive visual interfaces” (Thomas & Cook, 2005). In other words, it is a practice that combines our visual intelligence and data analysis techniques with interactive technology to get relevant information out of data. It allows data experts to walk through the analytical process with the non-data experts in the following three ways.

1. Visualizations enable us to see representations of the data that may not be easily understood just by looking at summary statistics

I am not proposing to replace statistics with visualizations (it is vitally important to look at the numbers), but to couple the delivery of summary statistics along with visualizations that show the shape of the data, it allows those who are less familiar with the data to understand what is going on. This allows the non-data experts to provide contextual input with ease.

A classic example of the value of data visualization is by Statistician Francis Anscombe. This table contains 4 bivariate data sets (4 lists of paired values). They are different, but almost all the summary statistics for each dataset are identical within 2 decimal places.
How are the datasets different? You can easily see that from the graphs of the four datasets.

2. Interactivity enables us to examine the data we need for the task at hand: “Analyze first, show the important, zoom, filter and analyze further, details on demand” (Keim, 2009)

Interactive visual interfaces allows users (both data experts and non-data experts) to switch between different tasks with nearly zero latency:

  • Looking at a summarized view or the “big picture”
  • Zooming into particular details (even individual records!) that are of interest by filtering or by drilling down
  • Examining relationships between different facets by linking visualizations, and have one act as a selector for another

Interactivity allows the user(s) to have a real-time “conversation” with the data — generate questions, make visual queries and get answers in real-time, (and come up with more questions. Rinse and repeat!). Interactive visual interfaces can be designed for different audiences — it can be designed for people who are not data experts, which makes the information much more accessible by non-data experts.

The video below is an example of wildlife strike analysis using an interactive visual interface (Tableau):

3. People trained in VA can talk to people on both sides

As mentioned earlier in this article, VA is a practice. In other words, it’s not only about the technology. Adopting VA tools may not be enough to solve the language gap problem, because ultimately a subject-matter expert or a decision maker would communicate with an analyst or data scientist — in essence, a person communicates with another person (at least for now).

Since collaborative analytics is an integral part of VA, people who are well-versed in VA come with the experience of talking to people with different skillsets and expertise, and can serve to bridge the data language gap between data experts and decision makers.

Note: In addition to VA experts, data experts with VA skills and decision makers with VA skills can work together to bridge the data language gap too.

This article was originally published on the Vancouver Institute for Visual Analytics’ blog. The Vancouver Institute for Visual Analytics trains students and professionals to not only become the deep analytical talent but also the bridge between data experts and non-data experts.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.