ISSUE 387  ·   May 17, 2022

In the News

Distilling the Laws of Physics From Raw Data

Researchers say we’re on the cusp of “GoPro physics,” where a camera can point at an event and an algorithm can identify the underlying physics equation. If you've studied genetic algorithms, the concepts here will feel familiar but data science is taking it to a whole new level.
Quanta Magazine | Charlie Wood

Organizations

Let's talk a bit about giving interviews

Advice is readily available for doing well in data science interviews when you're the one being interviewed. But interviews are completely different on the other side of the table. This is an insightful post for how to approach interviews when you're the one doing the interviewing.
Counting Stuff | Randy Au

Supercharge your Postgres for time-series data

TimescaleDB is the open-source relational database for time-series and analytics, built by developers for developers. It brings together the familiarity of PostgreSQL with speed and petabyte scale. Try it for free (no credit card required).

Tutorials, Projects & Opinions

Bandits for Recommender Systems

Bandits make great recommenders when new items are continually added (e.g., news, tweets) or when there's low traffic. While regular recsys tends to be greedy, bandits model uncertainty and deliberately explore. This is a great introduction to bandits with a focus on examples from industry. Includes linked references throughout.
Eugene Yan

Supervised Clustering: How to Use SHAP Values for Better Cluster Analysis

Cluster analysis is a popular method for identifying subgroups within a population, but the results are often challenging to interpret and action. Supervised clustering leverages SHAP values to identify better-separated clusters using a more structured representation of the data. Here's how the technique works, with clear examples along the way.
Aidan Cooper

Using Pyodide to Teach Data Science at Scale

Pyodide makes it possible to install and run Python packages in a browser. Pandas Tutor visualizes pandas chained method calls, step-by-step. Together, it's a browser-based data science education tool that easily scales. This post explores the use-case, the possibilities, and what it took to get Pyodide and Pandas Tutor to work together.
Pyodide blog | Sam Lau, Philip Guo

PostgresML

PostgresML is an end-to-end machine learning system that enables you to train models and make online predictions using only SQL. The goal is that anyone with a basic understanding of SQL should be able to build, deploy and maintain ML models in production.
GitHub | PostgresML

Transformers for Natural Language Processing

Don’t get left out in the cold stuck on one transformer platform. Learn how to use the right transformer, platform, and techniques to solve complex NLP problems with Denis Rothman’s new book, Transformers for Natural Language Processing, 2nd Edition. Available now on Amazon and the Packt website.
// sponsored

Resources

JavaScript for R

This new book shows how R and JavaScript can work together and it's much more than just JavaScript code running alongside R. This book shows how R and JavaScript can actively interact with each other and do things that aren't possible otherwise. Free to read online.
John Coene

Data Visualization

Since the critiques have started to roll in...

The New York Times published a grim data visualization over the weekend called "One Million Lost." It chronicles the million+ COVID deaths in the U.S. and it's a stunning portrayal of the pandemic. But is it effective? In this thread, Will Chase explores the challenges of visualizing death and what it accomplishes. 
Twitter | @W_R_Chase

How did you like this issue of Data Elixir?

👎    1     2     3     4     5   👍

Was this email forwarded to you?  Sign up here >>