Industrial data Visualization

A journey into Data visualization

Figure 1. Innovation deployment of a car factory.

What is data visualization

Files store values. Values are our data. A collection of data is information. Reading raw data stored in a file is possible. But interpreting raw data is impossible by reading it only. That is why we visualize them in an interpretable way.

“The greatest value of a picture is when it forces us to notice what we never expected to see.” -John Tukey

Today, we will drill down into Data visualization.

From a historical perspective, data visualization exists since a long time. According to “ Information Dashboard Design: The Effective Visual Communication of Data “ [1] in 1644, Michael Florent Van Langren, a Flemish astronomer, has provided the first visual representation of statistical data used to estimate distances between cities.

From a cognitive perspective, data visualization extends its roots to the 20th century. As cited in “ Cognitive Psychology “, [2] Max Wertheimer (1880–1943), Kurt Koffka (1886–1941), and Wolfgang Köhler (1887–1967) founded Gestalt psychology in the early 20th century.

Gestalt psychology is a field in psychology emphasizing that organisms ( we, humans !) perceive things in form of patterns or configurations rather than individual components. In other words, we tend to “group by” things when we visualize them. We might check out Gestalt principles for further information about this interesting subject.

The perspectives mentioned above explains to us that visualization seems to “ease” our understanding of data.

Let us see how we stand on this principle in visualizing industrial data.

Data visualization in Industries

As we might know, data is ubiquitous in the use of Data visualization in Industries. We have different kinds of data files storing voluminous data for different purposes.

In data visualization, we have one purpose: Taking decisions based on our visualized data.

It is crucial to define very well and understand the bases of Data visualization. We need to know what to visualize and why. Then we need to prepare our data. If we skip or don’t invest enough effort in this point, our visualization becomes difficult to read. Thus it prevents to reach our purpose mentioned above.

“To find signals in data, we must learn to reduce the noise — not just the noise that resides in the data, but also the noise that resides in us. It is nearly impossible for noisy minds to perceive anything but noise in data.” — Stephen Few

To visualize, we need to understand what content and context are.

Content is the quantity of something we want to measure. For example, the energy produced by a battery.

Content without context isn’t meaningful. We might have an in-depth look at what context in Industrial data visualization is. For example, the type of car we produce is our context.

Content and context are represented thanks to columns and rows in a data source. Rows and columns

Figure 2. Columns of a data source

Data in our data files are labeled with column names in Dashboards. Even for data contained in different sources like JSON objects, we can represent them as columns and rows.

In fact, columns can represent context or content. Once again, content is quantity whereas context is used to categorize quantity.

In our example above, we have content that could be the building type. and our content UnitPrice.

As we might guess, Dashboards are what come first to mind when we talk about visualizing industrial data. So what is a Dashboard ?

As mentioned in “Information Dashboard Design : The Effective Visual Communication of Data.”, a dashboard is [3] “A visual display of the most important information needed to achieve one or more objectives that have been consolidated on a single computer screen so it can be monitored and understood at a glance

Let’s stick to the definition above. Moreover, a Dashboard is a set of widgets.

So what are widgets?

Widgets are “parts” of a Dashboard containing a tool visualizing data. For example, a Bar chart is a tool used in a widget.

It is known coming from 1930’s English word gadget which is a small ingenious mechanical or electronic device or tool.

Presenting many widgets emphasizes the message we want to deliver to our audience. Indeed, only one widget might not be enough to take our decision. Also, every widget can highlight a specific behavior of our system.

Widgets are essential providing us a global vision on the “behavior” of data stored in a data source. This is life-saving for our industrial sector manipulating Big data.

In order to prepare a widget using a tool to display data, we need data!

So, how to prepare our data for widgets ?

Figure 3. Joining two tables in a data source to form a unique representation of data

In dashboarding, data source is a representation of data received from tables belonging to data files. We access them through Connections. We join data from these different Connections to get a main table that will be used by our widget.

In figure 3, we are joining two tables in a data source to form a unique visualization of the data extracted.

Figure 4. Result after joining the two tables.

When joining our context, we need to choose them wisely, otherwise, we might create meaningless data sources as shown in Figure 4. We can discuss what is joining columns in an upcoming article.

An important fact is that a widget uses only one data source but a data source can be used by different widgets.

So how do we prepare our connections ?

Figure 5. Connection types ( non-exhaustive )

In figure 5. we have a list of possible Connections we can use to import data.

We need to import our connections to our Dashboard design software and use them to create data sources.

Here we are, at one of the most delicate points of our topic. How do we choose our tools based on our data and our goal?

We should note that our goal is to communicate a message to our audience. A way of doing this effectively is [4] “ choosing an effective visual” as mentioned in “Storytelling with Data : A Data Visualization Guide for Business Professionals” . To have an insight on what context is, we might have a look at Context in Industrial data.

The decision flow chart below guides us in choosing the right chart for the right context:

We could check out briefly some widget examples and analyze their productivity.

Figure 7. Widget illustrating heat and power distribution in a factory.

In our figure above, colors are well chosen, labels and the layout are clear. Our data is properly visualized and catches our attention directly to heat which is a context. According to this Sankey visualization, the highlighted message is: “Our company uses mostly Natural gas to convert it into heat which supplies mostly our main building.”

Figure 8. Two widgets using Sankey and bar chart.

In figure 7, we have some data illustrated with our widgets. Our colors are clear. Yet compared to Figure 6, even if we have more widgets, the message is unclear. Our Sankey diagrams flows are like each other in terms of flow widths and widgets. Our context is unknown, apparently, our data had a poor naming! In our bar chart, our X and Y-Axis are unnamed and our widgets have poor titles. The only positive part of this example is that our bar chart demonstrates increasing content based on the chosen context. But as our context is unknown, it remains meaningless.

To summarize, we should pay attention to every point of our widgets.

“There is no such thing as information overload. There is only bad design.” — Edward Tufte

Coming to an end

Data visualization involves many fields from science to psychology.

We need to prepare our data before visualizing it.

As for our final words, we can say that data visualization in Industries provides an understanding of our industrial systems. It shows us ways to improve them.

References

[1] Few, Stephen. Information Dashboard Design : The Effective Visual Communication of Data. Sebastopol, Ca, O’reilly, 2006.

[2] Sternberg, Robert J.; Sternberg, Karin (2012). Cognitive Psychology (6th ed.). Belmont, Calif.: Cengage Learning. pp. 113–116. ISBN 978–1–133–31391–5.

[3] Few, Stephen. Information Dashboard Design : The Effective Visual Communication of Data. Sebastopol, Ca, O’reilly, 2006.

[4] Cole Nussbaumer Knaflic. Storytelling with Data : A Data Visualization Guide for Business Professionals. Hoboken, New Jersey, Wiley, 2015.

Knowledge enthusiast. Mostly about science, education and rhetoric. Writing about data viz in dashboards.