Yet another round of tech review for the past weeks. You have probably heard about it but this new post is also a way for me to close dozens of Safari bookmarks and archive some Twitter favs.

Want to get rid of Google services like me, e.g., Google Translate? Take a look at DeepL translator. It works like a charm when it comes to detecting the input language or for translation purpose as far as I can tell.

GitHub is an amazing place, not only for open source projects. If you are after some tech books for the week end, try making your bookshelf a little more wonderful each week (h/t @Jose_A_Alonso).

This monthly newsletter needs some Lisp dereferencing: Here is the Common Lisp homepage. And if you are in Clojure, Reducers, transducers and core.async in Clojure by Eli Bendersky provides a good overview of lazy sequences and trasnducers.

I’ve been satisfied with Python for almost 10 years. But I don’t think I’ll still be using it another decade from now. I think I’ll be using Swift.

Here is why. By the way, the Swift for TensorFlow project is already on GitHub.

There is a little update to Introduction to Python for Econometrics, Statistics and Data Analysis by Kevin Sheppard. If you are looking for a good introduction to numerical computation using Python 3.x, this is probably the best way to go. There is little about statistical modeling other than the linear model and some time series applications, but it is worth a read.

Understanding Machine Learning: From Theory to Algorithms, published in 2014 by Cambridge University Press, is available for free. What makes this book different from other books on the same topic? As most ML-related textbooks, it doesn’t start from probability or statistical distributions and the basis of regression and classification framework. Besides discussing several algorithms for ML models, it also features a 100-page long account of learning *per se*, including for instance a detailed overview of the PAC learning model, empirical risk minimization or MDL learning rules. On a related note, here is a recent post by Frank Harrell on those two cultures: statistical modeling and machine learning. Always a good read (see also My Journey From Frequentist to Bayesian Statistics), like Stephen Senn’s posts (e.g., Being a statistician means never having to say you are certain).

But, look, while the Mathematics for Machine Learning draft get updated recently, here is another little gem (which may end up being published at some point): Data Science, A Gentle Introduction (PDF, 310 pp.), by James Scott (h/t @ten_photos).

I must admit that I was really disappointed by the recent buzz around RMS, but I found *this* was very well put:

Of all the hills Stallman could choose to die on, he picked this one. 🤦♀️ https://t.co/4YZkZQ588X

— Jeff Atwood (@codinghorror) May 8, 2018

Hack is a typeface designed for source code. Should I try to replace Source Code or Fira Code? Not any time soon, but let us keep this in mind. Also, while VimR (not to be confused with the NVim-R plugin) has been refined and now use neovim instead of Macvim, I came across Oni, which is “a new kind of text editor.” Think of it as a combination of nvim powered by node. Don’t be afraid: There is also a Rust version of Emacs!

Have you heard of the Wes Anderson Palettes? There is an R package for that, but I would love to see one for Stata. Or you can visit the Rijksmuseum for free with the rijkspalette:

Here is an interesting post about the dat project by Tom MacWright: So you want to decentralize your website.

Other interesting posts or videos:

- Anatomy of a posit number, by John D Cook
- Metamorphosis and the Multilevel Model, by Richard McElreath
- Bayesian Non-parametric Models for Data Science using PyMC3, by Chris Fonnesbeck
- Programming R, and new features notably ALTREP, by Martin Maechler
- Seven Strategies for Optimizing your Numerical Python Code, by Jake VanderPlas

♪ Low • *Ones and Sixes*