Profile avatar
xangregg.bsky.social
Engineering Fellow at JMP, focused on #DataViz, preferring smoothers over fitted lines. Creator of JMP #GraphBuilder and #PackedBars chart type for high-cardinality Pareto data. #TieDye #LessIsMore
154 posts 1,522 followers 1,457 following
Regular Contributor
Active Commenter
Posts 19 Comments 31

Nice to see smoothers in Der Spiegel but I can't figure out what's going on with the "gewichteter Durchschnitt" trend line for BSW (magenta). It seems below almost all of its data points. Maybe some weighting I'm missing. Here's my smooth on scraped dots. www.spiegel.de/politik/deut...

submitted 14 days ago • 1 comment

Small #dataviz exercise: how would you communicate this college applicants data? www.dailytarheel.com/article/2025... year,rate,applicants 2019,22.63,42466 2020,23.54,44382 2021,19.24,53776 2022,16.85,57221 2023,18.74,57902 2024,15.56,66535

submitted 44 days ago • 5 comments

Counts of release year for songs played each month since 2011 on Radio Paradise. They're once again playing more new music, and the stripes around 1970, 2002, and 2006 are starting to fade.

submitted 57 days ago • 0 comments

Continuing my New Year's ritual of scraping Radio Paradise playlist data. Here's a bumps chart #dataviz ranking the number of times each artist was played by year. Radiohead still rules; sharp declines for U2 and REM in recent years; resurgences for Talking Heads and The Beatles.

submitted 58 days ago • 2 comments

Trying 6 jitter methods for my 2024 authentication codes. Rectangular grid, hexagonal grid and beeswarm, with and without smoothing (x-dimension jitter to reduce spikes). #dataviz

submitted 61 days ago • 1 comment

I logged all my two-digit authentication codes in 2024, after sensing more high-value codes in 2023. Maybe I was right. Low first digits are relatively sparse. #dataviz

submitted 61 days ago • 1 comment

"A highly simplified world map" #lessismore Source paper: faculty.collin.edu/dkatz/Rohde-...

submitted 66 days ago • 0 comments

Here's a better illustration of the difficulties of box plots with discrete data (Likert in this case, from poetry assessment study). In the second view the last group is clearly different from the others, but not so in the box plot version. #dataviz

submitted 70 days ago • 0 comments

I'm not fond of using box plots for discrete data--most look identical, but what's better? This AI poetry study has 7000 responses, so showing all the data isn't an option. Highest Density Region (HDR) plots aren't bad but aren't commonly known. But maybe there's just not much to see here.

submitted 71 days ago • 2 comments

📊 An old NOAA map of ❄️ probabilities is flying around socials that shows wonky color design Old 🗺️ (left): It takes work to understand because the color lightness doesn't increase consistently with values (48% is light-blue, 52% dark-purple 😬) New 🗺️ (right): lightness increases as values increase 👍

submitted 73 days ago • 3 comments

I think it was Howard Wainer who emphasized flattening curves to better visually understand variation (tho can't find a ref). Here's Sloane's Gap again as residuals from the trend curve. I think 10000 gets an extra boost because some sequences involve numbers in binary form.

submitted 73 days ago • 1 comment

Our paper "A Plot is Worth a Thousand Tests: Assessing Residual Diagnostics with the Lineup Protocol" has appeared in issue 3 of JCGS, www.tandfonline.com/doi/full/10.... . It is open access for now, but if you ever have problems accessing you'll find the preprint at doi.org/10.48550/arX...

submitted 102 days ago • 0 comments

I learned via Numberphile about Sloane's Gap, an empty band that appears when you plot counts of integers appearing in his Online Encyclopedia of Integer Sequences (OEIS.org). Here's the 2011 original and my updated version with 2024 data and coloring some categories. arxiv.org/pdf/1101.4470

submitted 78 days ago • 1 comment

Recent preprint, "Quantifying Data Distortion in Bar Graphs in Biological Research," looks to share some good bar chart metrics but seems far from peer review quality. www.biorxiv.org/content/10.1... Nonetheless it's been picked up by Nature media.nature.com/original/mag...

submitted 82 days ago • 1 comment

Another #dataviz from the ceiling height paper's data with an interesting pattern: density dot plot of the final course grades for two years of students, using hex grid jitter. 50 must be an important result at these Australian universities.

submitted 86 days ago • 0 comments

I can't stop looking at the data from this retracted university exam ceiling height paper. Unrelated to ceiling height, I found this interesting nonlinear fit for exam score versus coursework score. #dataviz www.sciencedirect.com/science/arti...

submitted 86 days ago • 2 comments

It was such a treat getting to visit Martha Wells and her lovely family and found family earlier this year. She’s well known now for her iconic Murderbot series, but Martha endured a number of heartbreaking trials over the course of her career as a working writer. She’s got some true grit.

submitted 97 days ago • 15 comments

Good thing I'm not in marketing. Here's a chart of registrations for my "Statistically Speaking" webinar earlier this month, with my bsky post marked. It's now on-demand: www.jmp.com/en_us/events...

submitted 98 days ago • 0 comments

Repost (sorry!) of “Great Validations: The Personal History, Adventures, Experiences, and Observations of a Visualization Researcher“, now with the right link: niklaselmqvist.medium.com/great-valida...

submitted 98 days ago • 0 comments