rogue7

10 ms·

1. 31.▲
   by rogue7 3y ago

   Finally ! Keyword-only arguments are the best to build reliable software. It's a bit like the redundancy of classes that led to attrs and dataclasses. It's the obvious next step.
2. 32.▲
   New Book: Model-Based Machine Learning by J. Winn and Chris Bishop et al.(twitter.com)

   1 points by rogue7 3y ago|0 comments
3. 33.▲
   by rogue7 3y ago

   I managed to install it with devbox (nix under the hood), works very well !
4. 34.▲
   by rogue7 3y ago

   In my line of work I've been using numpyro to model physical phenomena. There are multiple unbalanced categorical variables, so partial pooling helps a lot to infer the target in regions where the data is sparse.
5. 35.▲
   by rogue7 3y ago

   I've been using nix through devbox [0] & direnv recently for an elixir project. The experience has been good so far: - Fully reproducible. - Does not install things globally, rather on a per-project (directory) basis. - Reasonably
6. 36.▲
   A Probabilistic Betting Framework(blog.horaceg.xyz)

   2 points by rogue7 3y ago|0 comments
7. 37.▲
   by rogue7 4y ago

   Facebook has done it with their iOS Messenger app https://engineering.fb.com/2020/03/02/data-infrastructure/me...
8. 38.▲
   by rogue7 4y ago

   My thought exactly when I saw the domain name
9. 39.▲
   by rogue7 4y ago

   I just saw 2 roles when I clicked the link (Senior Front End/JavaScript Engineer | Senior Backend Engineer (3))
10. 40.▲
    by rogue7 4y ago

    I have made a small app regarding this topic: https://deaths.pages.dev
11. 41.▲
    by rogue7 4y ago

    I've thought about this and there's I think another reason coming from combinatorics. Consider a set of n cities. Being able to go from one city to another by plane only requires n airports. However, to have a complete rail networ
12. 42.▲
    by rogue7 4y ago

    How does mamba (and conda-forge) fare with respect to these metrics ?
13. 43.▲
    by rogue7 4y ago

    I've been using Numpyro for a while (2 years), and it's very fast, robust and well designed. I recommend the Statistical rethinking translation [0]. I've had some trouble with the plates sometimes, especially when using discr
14. 44.▲
    by rogue7 4y ago

    There is pandoc: https://pandoc.org/
15. 45.▲
    by rogue7 4y ago

    Hi, I am a data scientist, mostly interested in sustainability. I am proficient in data processing, probabilistic modelling, data visualization with 4+ years of experience. Location: Paris, France Remote: Possible Willing to relocate: No Te
16. 46.▲
    by rogue7 4y ago

    Looks very nice, how does it compare to arquero ? https://observablehq.com/@uwdata/introducing-arquero
17. 47.▲
    by rogue7 5y ago

    This is impressive, Copilot knows scikit-learn better than the data scientist that I am.
18. 48.▲
    by rogue7 5y ago

    Wow this is amazing. Now all I need to deploy static machine learning demos in documentation is https://github.com/pyodide/pyodide/issues/2198
19. 49.▲
    by rogue7 5y ago

    The approach JAX takes regarding PRNG for numerical computing is worth a look: https://github.com/google/jax/blob/main/docs/design_notes/pr...
20. 50.▲
    by rogue7 5y ago

    This book is very biased, see for instance the "one-hump world": https://www.jasonhickel.org/blog/2019/3/17/two-hump-world For a full review (in french): https://paulboosz.github.io&
21. 51.▲
    by rogue7 5y ago

    I usually use pandas' styled output to show tables in documents: https://pandas.pydata.org/pandas-docs/stable/user_guide/styl... Since it is programatically generated, I don't have to manually fiddl
22. 52.▲
    by rogue7 5y ago

    I agree with your comment, just FYI you could avoid apply on a pandas Series since it loops over every row in Python space. It's much faster to use vectorized operations directly e.g.: df["balance"].mul(36902.7).div(100e6) Se
23. 53.▲
    by rogue7 5y ago

    Wow I find this hugely impressive ! I'm a big fan of reactive notebooks such as Observable in ~javascript and Pluto in Julia. I've been thinking about something somewhat similar using js_of_ocaml instead of Rust+wasm, but I guess
24. 54.▲
    by rogue7 5y ago

    I believe you are looking for Gary Bernhardt's talk: https://www.destroyallsoftware.com/talks/the-birth-and-death...
25. 55.▲
    Concurrency in Python with FastAPI(horaceguy.pages.dev)

    1 points by rogue7 5y ago|0 comments
26. 56.▲
    by rogue7 5y ago

    Hi HN, Here are the basics about bets, i.e. things I wish I knew when I entered the field of sports trading. This is kind of an introduction to the field, with the formalism I developed on my own. Markdown source here[0] [0]: https:/&
27. 57.▲
    A Probabilistic Betting Framework(horaceguy.pages.dev)

    3 points by rogue7 5y ago|1 comments
28. 58.▲
    by rogue7 5y ago

    There are also these data points [0], that show that the newer generations have fewer and fewer wealth. [0] https://twitter.com/CharlotteAlter/status/143127950480736666...
29. 59.▲
    by rogue7 5y ago

    > is this a "things are getting better, stay the course" type video Yes. And sadly it isn't happening from what I can tell ; see for instance the latest IPCC report [0]. [0] https://www.ipcc.ch/report/
30. 60.▲
    by rogue7 5y ago

    At first I really liked Hans Rosling's book. But afterwards I read some very valid criticism [0]. > We see that while per capita income has indeed increased in the global South, the global North has captured the vast majority of new