Press "Enter" to skip to content

Category: Visualization

Visualizing Data over Time with F#

Published 2021-07-30 by Kevin Feasel

Codesuji takes us through creating an interesting video:

How is this accomplished? I reach into F#’s bag of tricks to leverage Deedle, Plotly.NET, and ffmpeg in order to transform a series of data files into a singular video showing county-level drought data from 1900-2016. Together these bring static data into a dynamic representation. For reference, the Palmer Drought Severity Index (PDSI) typically ranges from -10 (dry) to 10 (wet). Putting this all together is pretty straight-forward, but I wanted to call out a couple specific parts. For this particular example Deedle is overkill, but pairing it with Plotly.NET can often be useful in more complex situations. Plotly offers some nice customization options, which I take advantage of below. Once all the images are generated with Plotly, F# can shell out to ffmpeg to perform the video assembly. I do this in two parts, creating both an mp4 and webm file.

We’re reading datasets, parsing text files, deserializing JSON contents, building a visual for each point in time, and then creating a video out of it—all in 100 lines of code. Not bad.

Comments closed

Conditional Visibility in Power BI Paginated Reports

Published 2021-07-20 by Kevin Feasel

Sabrina Jordan has a clever solution to a common customer request:

How many times have you struggled to find the happy medium between a report that looks phenomenal when printed, but has everything a user might need in an Excel export? I recently built a beautiful paginated report with groupings separated by white space for easy readability – but the user wanted to export the results to Excel, and the format prevented them from sorting or filtering the report contents. Power BI Report Builder has a couple features that can allow you the best of both worlds, with a few simple tricks. By the end of this tutorial, you will have created two Tablix, set up conditional visibility based on report render format, and set conditional sizing on the Excel Tablix (using hidden charts!) to prevent blank pages.

Click through for the solution.

Comments closed

Font Choices with ggplot2

Published 2021-07-09 by Kevin Feasel

Kenneth Tay takes us through font options in R’s ggplot2 package:

I was recently asked to convert all the fonts in my ggplot2-generated figures for a paper to Times New Roman. It turns out that this is easy, but it brought up a whole host of questions that I don’t have the full answer to.

If you want to go all out with using custom fonts, I suggest looking into the extrafont and showtext packages. This post will focus on what you can do without importing additional packages.

A quick word of warning: R’s behavior with respect to fonts differs quite a bit between Windows and Mac/Linux. This becomes especially apparent if you do end up installing something like extrafont. H/T R-Bloggers.

Comments closed

Power BI Report Iconography

Published 2021-07-06 by Kevin Feasel

Joe Billingham brings in the Unicode characters:

There are 143,859 Unicode characters available, everything from emojis, symbols, shapes and braille patterns to dice and playing cards. Whether you want to offer further insight into your data, enhance the user experience or simply create something sublimely ridiculous, with so many icons at your fingertips, the possibilities are only limited by your imagination.

Click through to see how you can include specific Unicode characters to create a visual link in the mind of your viewer to your data.

Comments closed

When Bar Charts Don’t Start at Zero

Published 2021-06-30 by Kevin Feasel

Alex Velez has a confession:

Spoiler alert: I made a bar chart that does not have a zero baseline. I should add that this was intentional and not for teaching purposes, but rather because it seemed like the more appropriate solution for my given scenario.

This is an interesting edge case, but honestly, I think a bar chart is the wrong choice for the job here. I definitely prefer a dot plot here, even if the one Alex put together was a bit noisy. I think the best answer might have been to have a vertically aligned dot plot, similar to the final bar chart except without the bars. That reduces a lot of the noise while letting you label the specific points.

As far as the topic, I defer to noted philosopher Ron Swanson: “There’s only one thing I hate more than lying: skim milk. Which is water that’s lying about being milk.”

Comments closed

Using ggplot2 to Create a Faceted Histogram plus Curve

Published 2021-06-28 by Kevin Feasel

Sebastian Sauer builds a combo chart:

Overlaying a histogram (possibly facetted) is not something far fetched when analyzing data. Surprisingly, it appears (to the best of my knowledge) that there’s no comfortable out-of-the-box solution in ggplot2, although it can be of course achieved with some lines of code. Here’s my take.

Click through for Sebastian’s version, as well as information on the ggh4x library.

Comments closed

Pre-Sketching Data Visualizations

Published 2021-06-24 by Kevin Feasel

Laura Ellis explains the benefits of pre-sketching data visualizations:

When you take on a new data visualization project, it can be tempting to jump in and create visualizations right away with the idea that after enough exploring, the final format will present itself to you. And while it is important to dedicate time to EDA (exploratory data analysis), it can also be very beneficial to define a high-level plan early in the process.

Over time, I’ve found that producing an early sketch has been helpful in reducing the total amount of time and iterations taken towards building the end product.

Read on for the reasons why.

Comments closed

Color and Emotion

Published 2021-06-10 by Kevin Feasel

Cedric Scherer explains some of the psychology behind understanding of color in visuals:

Without any intention, the two variations of my visualization triggered different emotional reactions. While the red chart likely leads you to think “Wow, Berlin summers are quite hot,” the blue version may push you to think of summers as rainy and rather cold.

In general, we should have in mind that different details might spark different emotions and expectations in our viewers. Some of these details will make it easier for them to understand the chart in the manner the designer intended. 

Having experienced (parts of) a Berlin summer, let me confirm that they are not hot when compared to the midwestern or southeastern US.

Comments closed

Visualizing with Text

Published 2021-05-26 by Kevin Feasel

Alex Velez shows how you can use simple text to share information:

If you’re unsure about what I mean when I say simple text, I’m referring to the idea that just because you have numbers doesn’t mean you need to build a graph—as in this simple text example. Sometimes words with big numbers written in bold fonts (aka BANs) are more effective, especially when communicating one or two data points.

Some people’s hesitancy with simple text is that they think it’s unrealistic: the notion that you might only share one or two numbers with an audience. It’s is a fair point; it does seem a bit silly to think that you’d only talk through a couple of specific values when presenting data—even though I’m sure there are such occasions. Rather than thinking of the ideal use-case for simple text as when you only have one or two numbers in totality, consider when it may make sense to draw attention to one or two numbers in your larger story.

This is a good reminder that you don’t need everything to be fancy, shiny, and visual-laden. A little bit of text can go a long way in laying out a visual. That said, the warning is that text seems to be a little easier for people to miss, especially if there’s a lot of it. That’s where Alex’s explanation really pays off.

Comments closed