@Anopsy

need more data

 Amsterdam
Joined October 2013
  • Tweets2.9K
  • Following911
  • Followers197
308Media
Anopsy retweeted
MattyG@meawopplSep 10
Wife told me these wouldn’t fit. Little did she know I had trained for this moment for years.
549
20,167
815
362,351
12,013,304
can confirm
depths of wikipedia!
@depthsofwikiApr 21
53
Verrassende plaatjes. (Via reddit-dot-com)
21
29
2
299
17,479
Like "scheikunde" or even worse "natuurkunde", you think it's like "biology", but it's freaking "physics"
Wie Dan🍃@LudditeHackerMar 28
Prime example of "You can just do things". One guy, born a bastard, in 1548 in Bruges changed the Dutch languages forever from all its peers through his linguistic purism, inventing Germanic words for all sciences that use the Latin word in any other Germanic language.
68
Sprawa dla @Swietlan maps, woda u Ani Gotuje "bulgoni" i "mruga"?
1
36
Happy π day!
MATLAB@MATLABMar 14
10,000 digits of π visualized in MATLAB! Code ➡️ spr.ly/6017h4fjJ #PiDay
32
The oldest working planetarium in the world, Eise Eisinga Planetarium in Franeker, The Netherlands.
21
968
55
7,487
218,164
21
79
10
2,401
62,307
uv that's the tweet
3
1
11
638
Last year, I volunteered at PyCon DE & PyData—met amazing people, learned a ton, and got inspired. This year, I’m coming back as a speaker! 🤯 Excited to share insights from my internship at @TheRealNannyML and give back to the community that got me here. See you there! #PyConDE
1
169
Shapley values, originally from cooperative game theory, are now a powerful tool in #MachineLearning for feature selection and model explainability! 🤖📊 Check out my accessible #YouTube lecture, complete with linked demonstration code, to help you get started: youtu.be/oCIybnawLdg?si=5jvH… ∀.
12
81
4,110
For everyone that needs to hear it:
44
646
23
6,589
131,926
Some really nice words from my colleague 🥳 Thank you Santi
Santiago Víquez
@santiviquez12 Feb 2025
This post by @Anopsy is one of my favorites. One of those things I wish I had written haha It explains, how data drift, concept drift, and data quality affect an ML model's performance, in a fun and scientifically robust way. Everything is backed by cool experiments and interactive plots.
1
104
Backstreet Boys, Parmenides, and the degradation of ML model performance If you'd like to learn what can kill the performance of your ML model in production, check out my blog post where I explore the impact of missing values, data drift, and concept drift on ML models. 👇
1
1
102
nannyml.com/blog/ml-models-a…
1
48
Congrats to my colleagues Bialek and Kuberski for their research on confidence-based estimators for performance monitoring!
J. AI Research-JAIR@JAIR_Editor21 Jan 2025
New Article: "Confidence-based Estimators for Predictive Performance in Model Monitoring" by Kivimäki, Nurminen, Białek, and Kuberski jair.org/index.php/jair/arti…
2
78
uv pip install tensorflow in less than 10 seconds 🤯
82
New Narwhals Artwork 😍
1
80
Every time you say "garbage in, garbage out" an ML model dies.
1
1
4
297
🤣
Terrible Maps
@TerribleMaps5 Jan 2025
Fun fact: Poland hadn’t been discovered yet when Windows XP launched in 2001
1
87
Load more