Sign in

Wajdi

@wajdi.bsky.social
80 followers 217 following 184 posts

Statistician.. Data "Scientist" & Lecturer in #DataViz in France. Here to talk and share #datascience #DataViz #Python #math, #statistics and tech related stuff. www.WajdiBenSaad.com

PostsRepliesMedia
Wajdi @wajdi.bsky.social · 10/10/2026
Does multilingual text classification really need translation for every language? In my recent paper, I explored how selective translation and small language models can improve classification while keeping computational costs in check... medium.com/@wajdi.ben.s...
medium.com
Cost-Efficient Multilingual Text Classification
I have been spending the past couple of years working on a very interesting challenge set, one involving text from various sources…
000 0
Wajdi @wajdi.bsky.social · 29/09/2026
I'll be @neuripsconf.bsky.social in #Paris this year 🎉 my paper "Discovering Translation-Worthy Languages with E-Values" will be presented at the NeurIPS workshop "E-Values: From Statistics to ML" Greatful for my co-author Safa MADIOUNI and for her contribution and amazing work!
000 0
Reposted by Wajdi
4-Bit Daddy (After Dark) @xfourxbitxdaddy.bsky.social · 22/09/2026
Alright, so now my computer scans my bluesky, and maintains my "VIP Mutuals" list to keep me focused, auto-follows people that just popped up on my radar (unless banned), and auto-unfollows people that have gone radio silent (unless protected). Yeah, I'm a geek like that, old school and cyberpunk.
Just a bunch of nonsense.
5141 14
Reposted by Wajdi
GoldenViz @goldenviz.bsky.social · 05/08/2026
✨GoldenVizR v0.1.0 is now available on CRAN. review your ggplot2 visualizations against 25 rules covering completeness, readability, and integrity, and generate structured feedback and HTML reports. 🔽 install.packages("GoldenVizR") cran.r-project.org/package=Gold... #rstats #Dataviz #Tidytuesday
cran.r-project.org
GoldenVizR: Rule-Based Quality Checks for 'ggplot2' Visualizations Based on the 25 Golden Rules of Data Visualization
Quality-checking tools for reviewing R data visualizations built with the 'ggplot2' library against the 25 GoldenViz rules described at &lt;<a href="https://goldenviz.org/" target="_top">https://golde...
021 2
Reposted by Wajdi
GoldenViz @goldenviz.bsky.social · 24/07/2026
🔹A chart should represent data honestly without misleading the reader. 🔹The Integrity principle covers : - truthful scales, - baselines, - uncertainty, - outliers, - proportional areas, - data selection, - granularity, - and neutral framing. 🟣 Nine rules from GoldenViz. goldenviz.org
021 2
Reposted by Wajdi
GoldenViz @goldenviz.bsky.social · 24/07/2026
🔹 A chart should be readable without unnecessary effort. 🔹 The Readability principle addresses: - typography, - accessible colors, clutter, scales, formatting, chart choice, direct labels, and visual hierarchy. 🟣 Ten rules from the GoldenViz framework. goldenviz.org
121 2
Reposted by Wajdi
GoldenViz @goldenviz.bsky.social · 24/07/2026
🔹 A chart should give readers everything they need to interpret it 🔹 The Completeness principle covers: -clear titles, - labelled axes, - missing data, - legends, - sources, - context, - and useful annotations. 🟣 Six rules from the 25 Golden Rules of Data Visualization goldenviz.org
111 1
Wajdi @wajdi.bsky.social · 22/07/2026
humm...this is an edge case with the current implementation.. GoldenVizR checks scale truthfulness through bar, column, and area baselines, but does not yet inspect scale_fill_gradient2() midpoints , so It returns a false PASS here :( but it's a great test case for the next version..!
020 2
Wajdi @wajdi.bsky.social · 22/07/2026
It took a (long) time but it's finaly out! an #rstats adaptation of the 25 golden rules of data visualization is out ! Feedbacks and comment are welcome!
050 5
Wajdi @wajdi.bsky.social · 17/07/2026
original post: www.reddit.com/r/mildlyinte...
reddit.com
From the mildlyinteresting community on Reddit: Someone left this note encouraging people to switch to Linux in our mailbox
Explore this post and more from the mildlyinteresting community
010 1
Wajdi @wajdi.bsky.social · 17/07/2026
I've just seen this posted on reddit.. someone put it in their mailbox, encouraging people to switch to #Linux.. and I find it awesome! instead of getting marketing crappy flyers and catalogues, I'd rather open my mailbox to find this ! #OSS #OpenSource #LibreOffice #Flathub
160 6
Wajdi @wajdi.bsky.social · 16/07/2026
i hope it is the last time records are broken…
020 2
Wajdi @wajdi.bsky.social · 15/07/2026
this look cool! the images of different species is a nice touch 👏
010 1
Reposted by Wajdi
Georgios Karamanis @karaman.is · 15/07/2026
This week's #TidyTuesday is about the penguins in the AVONET database of bird traits (Tobias et al. 2022). Ten body measurements for 93 penguins, across 18 species and six genera Code: github.com/gkaramanis/t... #RStats #dataviz
A grid of small ridgeline plots titled "Measuring penguins". It compares ten body measurements across six penguin genera. The genera run down the rows: Aptenodytes, Eudyptes, Eudyptula, Megadyptes, Pygoscelis and Spheniscus. The ten measurements run across the columns: beak depth, beak length at the culmen and at the nares, beak width, first secondary length, hand-wing index, Kipp's distance, tail length, tarsus length and wing length. In each panel, three smoothed curves show how the measurements are distributed by sex, with females in red, males in orange and unknowns in blue, and a small triangle marks each penguin. Eudyptes and Pygoscelis have the fullest data across all three groups. Eudyptula, Megadyptes and Spheniscus have much less, often just a few birds of unknown sex. Aptenodytes, the largest penguins, sit furthest to the right on wing and tail length.
0121 12
Wajdi @wajdi.bsky.social · 14/07/2026
and the code : github.com/WajdiBenSaad...
github.com
TidyTuesday/2026/2026-07-14 at main · WajdiBenSaad/TidyTuesday
Repo for the #TidyTuesday challenges. Contribute to WajdiBenSaad/TidyTuesday development by creating an account on GitHub.
000 0
Wajdi @wajdi.bsky.social · 14/07/2026
here is this week’s #TidyTuesday visualization! 🐧 I explored how ten anatomical measurements vary across six penguin genera, combining individual specimens with medians and interquartile ranges. I gave it a light Antarctic field-atlas look as much as i could 😅 #dataViz #rstats
150 5
Wajdi @wajdi.bsky.social · 14/07/2026
🎮 also made an interactive version of the chart Hovering over each pixel reveals the film, release date, scores, director and game publisher, while the level cards filter the films by decade.. Here’s a quick preview of how it works. 🎮 #TidyTuesday #RStats #DataViz
010 1
Wajdi @wajdi.bsky.social · 14/07/2026
code here for the ggplot and for the interactive version : github.com/WajdiBenSaad...
github.com
TidyTuesday/2026/2026-06-09 at main · WajdiBenSaad/TidyTuesday
Repo for the #TidyTuesday challenges. Contribute to WajdiBenSaad/TidyTuesday development by creating an account on GitHub.
000 0
Wajdi @wajdi.bsky.social · 14/07/2026
Still trying to catch-up! I finally finished my #TidyTuesday #Rstats #DataViz on "Films based on video games" 🎮📊 So for this one, I wanted to see whether these adaptations have actually improved over time. the median Rotten Tomatoes score rose from 17 in the 1990s to 51 in the 2020s
150 5
Wajdi @wajdi.bsky.social · 10/07/2026
and here’s how the visualization came together, from a regular heatmap to the final UFC-inspired design... I gradually added the octagons, red palette, dark background, cage texture, typography, logo, and finishing details... #TidyTuesday #RStats #DataViz
091 9
Wajdi @wajdi.bsky.social · 10/07/2026
code is here : github.com/WajdiBenSaad...
github.com
TidyTuesday/2026/2026-07-07 at main · WajdiBenSaad/TidyTuesday
Repo for the #TidyTuesday challenges. Contribute to WajdiBenSaad/TidyTuesday development by creating an account on GitHub.
010 1
Wajdi @wajdi.bsky.social · 10/07/2026
I finally managed to finish this week’s #TidyTuesday visualization! the topic is something I’m familiar with, but it still took me a few hours and many iterations to make the chart look decent. Here it is: "INSIDE THE FINAL ROUND" : how UFC fights end across five rounds #RStats #DataViz
140 4
Wajdi @wajdi.bsky.social · 09/07/2026
try with one of these libs in the blogpost ( disclaimer: i wrote this article) medium.com/@wajdi.ben.s...
medium.com
The Complete Guide to Python Data Visualization Libraries
Python has become the go-to language for data visualization, and for good reason. Its ecosystem offers a tool for every situation, whether…
020 2
Wajdi @wajdi.bsky.social · 09/07/2026
if AI does not help get you there in around an hour of prompting, might as well plug the final prep'ed dataset to a reporting tool ( if you have any) and give them the standard output, and move on to something else. if there is no reporting tool, and matplot lib is not getting you there, 1/2
100 1
Wajdi @wajdi.bsky.social · 09/07/2026
this one is for sure a bot 😂
static.klipy.com
Emmanuel Macron Emmanuel Macron For Sure
ALT: Emmanuel Macron Emmanuel Macron For Sure
000 0
Wajdi @wajdi.bsky.social · 09/07/2026
exactly , my first thought... I just don't know what's the use of doing this eventually? what's the catch 🤔
100 1
Wajdi @wajdi.bsky.social · 09/07/2026
the Tee looks cool!, happy 20th anniversary :)
210 2
Reposted by Wajdi
LibreOffice @libreoffice.bsky.social · 08/07/2026
Our next major update is coming! #LibreOffice 26.8 will be released in August, with many new features, improvements and fixes. Help our community to test the first beta version: qa.blog.documentfoundation.org/2026/07/08/l... #foss #openSource #freesoftware
LibreOffice 26.8 banner
7538101 538
Wajdi @wajdi.bsky.social · 09/07/2026
2- do data prep in python until the last bit, and connect a proper reporting tool, either a paid one ( power bi, tableau, qlik..etc). but this would come at the cost of not being able to fully customize it to the last bit... and sometimes, customizing reporting happens in iterations..
010 1
Wajdi @wajdi.bsky.social · 09/07/2026
1- invest a lot of time customizing the template of the charts/report, this way i schedule it and forget about it, unless something breaks or no data or data issues, so report is empty.
110 1
Wajdi @wajdi.bsky.social · 09/07/2026
may i ask why it has to go through python? If i get this correctly, you have a data pipeline on python, and would like to automate it end to end to the reporting outputs? from what i have been able to do previously, i had 2 choices for such case:
200 2
Wajdi @wajdi.bsky.social · 08/07/2026
if you are trying to do a very customized infographic-like viz, I think the best route is to either: use and be assisted by an AI coding agent or change the tool to GIMP or Illustrator ...
110 1
Reposted by Wajdi
Compute! Paris @computeparis.bsky.social · 07/07/2026
At Compute! we celebrate people who produce code that is accessible to all and maintain the software hundreds of us rely on every day. They do work most users never see, and the result is infrastructure anyone can build on, inspect, and improve.
111 1
Wajdi @wajdi.bsky.social · 08/07/2026
and here’s how it came to life. This one took me longer than most plots, but I wanted the final map to feel thematic and to honor the legacy, languages, and heritage of the African continent... #TidyTuesday #DataViz #RStats #ggplot2 #Maps
0120 12
Wajdi @wajdi.bsky.social · 08/07/2026
Code is here: github.com/WajdiBenSaad...
github.com
TidyTuesday/2026/2026-01-13 at main · WajdiBenSaad/TidyTuesday
Repo for the #TidyTuesday challenges. Contribute to WajdiBenSaad/TidyTuesday development by creating an account on GitHub.
000 0
Wajdi @wajdi.bsky.social · 08/07/2026
Catching up on a January #TidyTuesday dataset: The Languages of Africa. I turned the continent into a hex-tessellated map, with each country split by language-family composition, sand texture & scattered letters background & borders to echo the theme visually #DataViz #RStats #ggplot2 #Maps
190 9
Reposted by Wajdi
Manish Datt @manishdatt.com · 07/07/2026
AI-assisted categorization of Nicknames. Dashboard: dataviz.manishdatt.com/posts/2671 #TidyTuesday #dataviz #Javascript
AI-assisted categorization of Nicknames from the UFC athletes and fight data.
161 6
Wajdi @wajdi.bsky.social · 07/07/2026
this is a very cool idea !
010 1
Reposted by Wajdi
Denis Mongin @denismongin.bsky.social · 07/07/2026
ça y est. 2026 est passée 2ieme en sévérité cumulée de vague de chaleur depuis 1950. @sergezaka.bsky.social Quelques #dataviz depuis les données @infoclimat.fr pour illustrer cette folie ⬇️🧵 #1: sévérité cumulée des vagues de chaleur sur l'année. 2023 est encore devant, mais pas pour longtemps
Diagramme en barres horizontales : vagues de chaleur en France de 1959 à 2026, années classées par sévérité cumulée (somme des °C-jours au-dessus de 23,4°C). 2003 domine très largement, à environ 73 °C-jours. 2026 arrive 2e, à environ 57, devant 2022 et 2025 (~54). Suivent 2019 et 2006 (~36–38), puis une longue traîne d'années plus modérées jusqu'à 1959 (~4) tout en bas. Chaque barre est segmentée et étiquetée par mois (juin, juillet, août), indiquant quand la sévérité s'est accumulée. Les années records 2003, 2026 et 1976 sont en rouge.
1812 12
Wajdi @wajdi.bsky.social · 07/07/2026
super analyse, franchement très clean 👏 ça ne tombe pas dans le délire du pic spectaculaire : on voit le vrai sujet, le truc qui dure, qui s’accumule, qui pèse.... ça donne une lecture beaucoup plus juste de la vague de chaleur. gros taf... bravo @denismongin.bsky.social
221 2
Wajdi @wajdi.bsky.social · 07/07/2026
to sign-off, i'd say the missing ingredient here is not just /only a better scale/chart type... it’s a pinch of integrity...
000 0
Wajdi @wajdi.bsky.social · 07/07/2026
the key is that the visual drama should match the real-world impact... in this case, the issue is not only the axis choice; it is that the chart makes a small felt difference look like a major event... and would make the headline, cause a much more higher "psychological impact".. etc
100 1
Wajdi @wajdi.bsky.social · 07/07/2026
so i’d rather use a line chart, dot plot, or annotated range, with a contextual scale, but with the 2°F difference explicitly labelled...
100 1
Wajdi @wajdi.bsky.social · 07/07/2026
but i don’t think that is the most useful design here either, because the real question is not the total temperature magnitude. it is the day-to-day variation and whether that variation is meaningful...
100 1
Wajdi @wajdi.bsky.social · 07/07/2026
well, for this specific situation, ie. weather charts and analyzing temperature evolution over a week, i would probably avoid bars altogether... bars encode magnitude from a baseline, so if we keep bars, i would start them at 0°F...
100 1
Wajdi @wajdi.bsky.social · 07/07/2026
by the way, im a big fan of your work 🫡
110 1
Wajdi @wajdi.bsky.social · 07/07/2026
the rule I mentioned is not just about “starting at zero” or “choosing the right axis”; it’s about making the scale honest for the context. So : either the chart type should change, or the scale should start at zero. but of course, that would not serve the fear-feeding agenda...
200 2
Wajdi @wajdi.bsky.social · 07/07/2026
here, the issue is that a 2°F difference is visually dramatized far beyond its real-world impact or felt difference... the problem is not only the scale; it’s the mismatch between the visual drama and the real-world meaning
100 1
Wajdi @wajdi.bsky.social · 07/07/2026
I think they need a course on data viz rules.. I would recommend the 25 Golden Rules of Data Viz.. but hey, if i only have one rule, it would be : Rule 17: Appropriate scale goldenviz.org/en/integrity...
100 1
Wajdi @wajdi.bsky.social · 06/07/2026
at the Academy of Athens.. few days ago a blend of ancient inspiration and neoclassical splendor... #Greece 🇬🇷
010 1