Press "Enter" to skip to content

Category: Visualization

Choosing Colors for Visuals

Mike Cisneros has some guidance around color selection when designing visuals:

Regardless of how you select it, this key color will be used to denote the data points, or the data series, on which you feel it is the most important for your audience to focus.

All of the other colors we use will be based on where they are on the color wheel in relation to this key color, how many colors we intend to use, and what kind of relationship the rest of the data has to the data represented by the key color. 

I’ll admit that I just reach for the pre-created palettes which have done the work for me already.

Comments closed

The Value of Table Visuals

Shannon Holck gives us several scenarios in which tables are a good choice of visual:

Exact numbers are needed
If your report goes to a customer and you need an exact quantity or price or total, absolutely, a table is likely needed.

Displaying a few discrete values
If you need to display 5-10 things and they all represent unique values, a table may be appropriate.

Need to establish trust in the data
If you don’t trust the data (yet) and want to verify data, you can sometimes compare data at a roll-up level to a known source.  This is a great way to test not only the data but that the calculations are accurate.  

Click through for more reasons, followed by cases in which you want to avoid table visuals, and finally a few ways to improve tables. I’m not the biggest fan of the improved tables (because I want simpler and denser), but this is good food for thought.

Comments closed

Visualizing Ranking Data

Stephanie Evergreen gives us a few techniques for visualizing ranking data:

And any time your data could be visualized in a bar chart, you can always take a jump to a dot plot or lollipop chart. You got this.

Any of these variations will be a perfectly fine visual to show rank data at a single point in time. If you have rank over time OR rank comparison across multiple groups, try a Bump Chart.

I was going to recommend a Cleveland dot plot, myself.

Comments closed

Color Band by Group in Power BI

Marco Russo and Alberto Ferrari show how we can change color alteration to switch from row to row and instead go from group to group:

The background color of the rows depends on Sales[Order Number]. The background color switches between white and light gray every time the order number changes, so all the rows of the same order have the same background color and can be easily identified. You cannot obtain this visualization by only using a Power BI style, because the coloring of a row depends on the actual data in it. You can achieve this goal by using the conditional formatting feature in Power BI. You can set the background color of a cell according to the value of a measure. Therefore, you need a DAX formula that returns two values: one for the white rows and one for the gray rows. The value returned by the measure must alternate between those two values with each consecutive order number.

Read on for an example of how you can do this.

Comments closed

Progressive Disclosure in Power BI

Prathy Kamasani takes us through the implementation of a design idea in Power BI:

In the above example, I used a pattern to show details using action from the Card. When a user clicks on a card, the report will show details related to Card. It sounds straightforward, but it involves a lot of work using Power BI Functionalities: Buttons, Bookmarks, Sections, Grouping and Page Size.

There are few aesthetics I paid attention in this Report Page which are key for any landing page. Usually, a Landing page helps users to navigate around the Power BI Model, so it is important to highlight those navigation steps. In the above model, I used Buttons, labels and Images for navigation hints.

I like this for some uses, like giving analysts a chance to dive into the data. For an operational dashboard, I don’t like it very much unless the cards at the top alone provide me enough information to know whether I need to take an action; otherwise, it loses one of the most important concepts of a dashboard, glanceability.

Comments closed

Avoiding Diagnonal Axis Labels

Cole Nussbaumer Knaflic gives us two good alternatives for avoiding diagonal labels in data visualizations:

There is one common phenomenon in graphs that I recommend actively avoiding: diagonal axis labels. They are often observed on the x-axes of graphs, where many tools automatically rotate text when the labels become too long to fit horizontally. While this might seem like a kind favor, there are usually better options. Beyond looking messy, diagonally rotated text is slower to read. In this short post, I’ll highlight two common scenarios that lead to diagonal x-axis labels—long category names on bar charts and long date labels on line graphs—and a couple ideas to try instead.

Diagonal labels aren’t the worst on printed visuals (as you can tilt the paper to read those labels clearly), but they’re not great. When combined with screens—especially screens which change their rotation as you tilt them, like on phones—that leads to a lot of unnecessary dissatisfaction.

Comments closed

Building FAQs on a Power BI Dashboard

Evan Rhodes takes us through building out a FAQ for a dashboard:

I suppose you could add a bunch of text boxes with questions and answers. But, what if you have several questions and there isn’t enough space? I’m reminded of something a fantastic boss once told me, “Never pass up an opportunity to wow someone and grab their attention with your work.” So, let’s add some wow effect to this by leveraging the bookmarks and buttons functionality.

Bookmarks and buttons allow us to create a user experience that is intuitive to the user and that allows them to navigate around the page easily by just clicking. In this case, click on a FAQ and the answer appears. Click on the FAQ again or a different FAQ… I think you get the point.

If you need this on the dashboard itself, this is probably the right way to do it—there for the one time you need it and hidden away the rest of the time.

Comments closed

The Importance of Gridlines

Stephanie Evergreen shows why (subtle) gridlines are so important in visuals:

Here’s the thing: This chart NEEDS gridlines. I’ve said this before but I find this anti-gridline trend so common that I need to address this topic explicitly.

The *medium gray not black* gridlines are necessary because I do not have data labels on every one of the dots in the chart. 

A quick reminder is that even Edward Tufte (a key proponent of the “gridlines are bad” school) doesn’t hate all gridlines. Subtlety is key with them: they should be there when you need them but easily ignored when you don’t.

Comments closed

Changing the Graphics Device in RMarkdown Docs

Colin Gillespie shows us how to change PDF and PNG output settings within knitr:

In many workflows, function calls to graphic devices are not explicit. Instead, the call is made by another package, such as knitr.

When kniting an Rmarkdown document, the default graphics device when creating PDF documents is grDevices::pdf() and for HTML documents it’s grDevices::png(). As we demostrated, these are the worst possible choices!

Click through to see what you can do about it.

Comments closed

Saving Graphics in R Across Multiple OSes

Colin Gillesipie takes us through exporting graphics in R and some of the cross-platform foibles you’ll find:

One of R’s outstanding features is that it is cross platform. You write R code and it magically works under Linux, Windows and Mac. Indeed, the above the code “runs” under all three operating systems. But does it produce the same graphic under each platform? Spoiler! None of the above functions produce identical output across OS’s. So for “same”, I going to take a lax view and I just want figures that look the same.

Read on to understand the differences and hopefully limit confusion around them.

Comments closed