2019
I like minimal theme, hence the Hugo them I choose last year. However, I just found an even more minimalist theme: slim.
Scott’s World*, and more animations to see on Complexity Explorables.
Even if I have only increased the length of my working days by 2 hours in 1
month (currently 9am to 3pm), I definitely stay out of work for a good 2 or 3
hours once I get home. I guess I just have to live with that for the moment.
It’s probably time to finish Occupied before the beer finishes me off. #self
Functional programming explained for the pragmatic programmer. Nice take. Maybe it would have been easier to focus on C versus Common Lisp before addressing the case of hybrid languages. (via HN)
What a beautiful artistic work at the crossroads between dataviz and
infographics, by @janezhgw. #dataviz
Here is the fourth edition of Algorithms, by Sedgewick & Wayne, a definitive book to have after Knuth’s monumental work and the Cormen et al. (via @TechSparx)
Scientists rise up against statistical significance. Together with Moving to a World Beyond “p < 0.05”, it is probably time to rethink statistical significance and embrace the world of uncertainty instead. As Stephen Seen once said:
We can predict nothing with certainty but we can predict how uncertain our predictions will be, on average that is. Statistics is the science that tells us how.
An Introduction to Applied Bioinformatics: An interesting online textbook that I
found while browsing the scikit-bio Python package on Github. #python
Interesting to know: The wakefield R packages allows to quickly generate random
data sets. I learned about that while reading David Gohel’s Using R as a BI
tool. #rstats
On the simplicity of working with a Terminal: processing 44K of mails in less than 2 seconds.
Too late to start re-reading Don Knuth’s excellent book on Mathematical Writing (PDF), but I will definitively do it in a few days.
What is Data Science after all? I never liked this term, and I consider myself as a statistician, or better a data craftsman, because I mostly spend my time dealing with data after all. Stephanie C. Hicks & Roger D. Peng wrote a nice article, Elements and Principles of Data Analysis, which I believe provides quite an honest account of DS-related stuff:
Data science is the science and design of (1) actively creating a question to inves- tigate a hypothesis with data, (2) connecting that question with the collection of appro- priate data and the application of appropriate methods, algorithms, computational tools or languages in a data analysis, and (3) communicating and making decisions based on new or already established knowledge derived from the data and data analysis.
I was just reading some of Rackhim’s posts. He’s the author of the recent EmacsCast. The one on backups is quite interesting. I use Arq (Thx @fonnesbeck!) daily since 5 years or so and I have been happy with that only one solution to backup my personal and work-related data. I no longer use cloud fronts like Dropbox, except for already anonymised stuff I don’t get care enough to bother with privacy.
Foundations of Machine Learning. Never heard of it before I spotted @gappy3000 tweet.
Moving to a World Beyond “p < 0.05”. Or maybe the earth isn’t just round. (via @kaz_yos)
Sudoku solver written in more or less 30 lines of Racket code. #scheme
TIL There are several flavours of awk lurking around on the internet. Here is
bioawk, a bioinformatics-aware awk program.
📖 Alberto Moravia, L’amour conjugal (Denoël, 1948)
Performance of Error Estimators for Classification (PDF). Always good to be remembered of how important error estimation is in statistical modeling, especially with small samples. Remember Frank Harrell’s post?
Slate “helps you create beautiful, intelligent, responsive API documentation.” It reminds me of the whole stack of racco (probably dead), docco (sill live) & Co. Slate is used in Clojure by Example, a site that offers an original and very instructive approach to learning the basics of the Clojure language.
📖 Delphine de Vigan, Les gratitudes (JC Lattès, 2019)
I have been seriously thinking of subscribing to NordVPN during the last few months. On further inspection, there was a good deal for the 3-year subscription plan. Now, it’s done.
Essential Statistics with Python and R. Although this textbook does not cover advanced material (and the figures are terrible), it comes with a lot of exercices that one can solve using either R or Python.
Flux ML and differentiable programming. Nice to see how new packages are
continuously coming in the Julia ecosystem, after so many years. #julia
TIL DuckDuckGo, which has been my default search engine since 2018, features a built-in URL shortener. So nice! (via Brett Tersptra)
The number of projects hosted under the Apache Software Foundation never ceases to fascinate me. Today, I discovered Jena for the semantic web!
The Definitive Guide To Syntax Highlighting. Nice to see some good old posts
about Emacs. It makes me want to activate the paren-face mode to change a
little. #emacs
Explorable multiverse analyses. What a talent this guy has! (via @mjskay)
It looks like Statistical Rethinking will have a profund impact on bayesian
statistical computing. There’s now a Julia package to complement the R one. (via
@zerology) #julia
According to BSAG, Doom Emacs has been polished a little in recent months. I’m
still on Spacemacs–probably for a long time to come–but I remember how
pleasant the experience with Doom Emacs was. #emacs
Mathematics for Machine Learning is finally out. (via @ChengSoonOng)
RMS is now taking care of Apple. Now, I can’t help but smile at the idea of this
picture where we see RMS carrying his laptop on his shoulder. Surely he wasn’t
listening to music on iTunes. Note too that the list of criticisms made of
Microsoft is much shorter (fair enough), but the same is true for Google who
only gets two dozzns of bad marks! #apple
Commit Often, Perfect Later, Publish Once. This reminds me of Stack Overflow motto circa 2010 (“Vote early, vote often”). Anyway, this recommended best practices with Git are very well done.
Don’t let tomorrow’s beauty stop you from performing continuous commits today.
How about generating figure name using MD5 hash? I’ve long been wondering how to
store unique file names for all documents that I happen to write from day to
day. The last few years, I decided to prefix all such file names using either
fig- or img- depending on the context (i.e., whether it has been generated by a
computer program or in the case it’s just an illustration grabed on the
internet), followed by a short but meaningful description, e.g.
img-emacs-screenshot.png. When it is a series of figures, I usually append an
index (“a”, “b”, …; or zero-padded numbers). Still I have lot of duplicates
file names on my HD. One way to circumvent this issue is to generate random
hash, or I believe so since we all have the md5 utility on Un*x systems. Here we
go:
current master ✗ 74db262 22h25m ✖ △ ◒ md5 -s "emacs-screenshot"
MD5 ("emacs-screenshot") = 65c9ef7d939db96dd290adcf9597d65b
Small Sharp Software Tools. Together with Vince Buffalo’s Bioinformatics Data Skills, I believe this combo should provide the very best technical exposition to practical Unix. You may want to add Learning Unix for OS X if you’re interested in Mac-specific tools. (Disclaimer: I haven’t read Hogan’s book yet).
Time to watch The Expanse, Season 2, now.
The lsp backend for the Python layer in Spacemacs has so much improved over
time, and it is much more featured than the default anaconda one. Pending minor
issues with mypy which complains about missing imports (this can be resolved
using a config file, as described here), everything works perfectly. Things are
going too fast for me with the develop branch of Spacemacs. #emacs
A (very) short intro to Constraints: Nice visual explanation à la idyll. (via @JohnSelstad)
Learning Statistics with R. Looks like a nice intro to statistics with R. I
personally started with Peter Dalgaard’s Introductory Statistics with R, but no
doubt this should be a good start too (beware this tutorial relies on external
packages). #rstats