Skip to content

DSB #122


the weekend is near, DSB also, and with it so many good articles. I love the article about the Battleship game from Analytical. I don’t like the message of the article about possible EU restrictions on clickstream data from Pop.

No matter what, as always, enjoy your reading.

Analytical – We all know Battleship, a game where you try to estimate the position of the enemy ship and sink the entire fleet. Now you can play it with the help of Bayesian search theory. (rcmd by reader) – Don’t be afraid of heuristics, sometimes it’s more suitable than ML. – At Canva they do distributed hyperparameter optimization with Argo Workflows.

Computer Science & Science – Kritická reakce na video why you shouldn’t learn Python in 2021. (rcmd by reader) – Sim2Real techniques – train Robot in simulation first and then let it live in the real world. – Python 3.10 comes with structural pattern matching. What is it and when/how to use it?

Graphs and Visualizations – ggHoriPlot makes your life much easier if you want to create a horizon plot in ggplot in R. – Have you ever heard about Bokeh, an interactive visualization library in Python? Now you have. – Cutecharts
in Python are perfect for funny looking hand made charts.

Business and Career – Another neobank is alive. Let me introduce you
Chase by JPMorgan. – Comprehensive survey about salaries in Data Science. So many details like which library earns you the most money and many more. – Are you a tech worker? Then maybe you should read this article on how to behave on the job market.

Pop – EU will maybe forbid using search queries and internet browsing history for the modeling of credit scores. Some other personal data will also be prohibited. (rcmd by reader) – End-to-end business intelligence is dead, what now? – Dimensions are difficult but fascinating, how do mathematicians handle them?

Education – Long time no see, let’s meet again with quantile regression. (rcmd by reader) – Learn about mutliple sampling techniques. – Nice example on how to build in Django. In this case you are working on a personal diary.

Data & Libraries – Make your CMD beautiful with Cleo! (rcmd by reader) – Is programming an art? Then you need to practice in order to master it. PyEtudes will help you to find proper material to train yourself. And no, these are not online courses, but proper programming exercises. (rcmd by reader) – List of tools for data labeling. Because you don’t want to have bad labels.

MLOps – How to
build custom MLOps. – An end-to-end data science project can be too much for only one data scientist when the abstraction and the infrastructure are poor. – You need data engineering skills to be a good data scientist.

Video & Podcast Oliphant, creator of NumPy, SciPy, and Anaconda in Lex Friedman podcast. (rcmd by reader)

Papers & Books – Overview of overparameterized models in ML. – Scaling TensorFlow predictions.

Behind the Fence – Senior Data Scientist in TikTok in LA, USA.

Joke Simpson’s Simpson’s paradox (rcmd by reader)

Be First to Comment

Leave a Reply