ISSUE 383  ·   April 19, 2022

Insight

Resist the Urge to be Impressed

It's easy to be impressed by technologies like GPT-3 but we're far from an autonomous AI that could live up to the prevailing media hype. In this post, Emily Bender offers a realistic perspective on what the media gets wrong about AI.
Emily M. Bender

Organization

Data team structure: embedded or centralised?

There's an ongoing debate over whether data teams should be centralized or embedded within departments. Getting the data team structure right is no small deal as a lot of things will fall in or out of place depending on how you set your team up. Here are some key considerations and ways to approach them.
Mikkel Dengsøe

Thinking about evolving your data stack

There are a lot of articles about the modern data stack and where things are going but ultimately, what do you really need? Here are some tips for thinking through the options when modernizing your data stack.
Emily Thompson

Sponsored Link

Introduction to Automated Data Labeling

Preparing high quality vision datasets is a long and cumbersome process. This is why companies are now using automation to accelerate labeling and QA. Take a dive into automating data preparation with Superb AI and whether or not automation is right for you and your team. 

Tutorials, Projects & Opinions

The Dunning-Kruger Effect is Autocorrelation

Great post explaining the Dunning-Kruger effect, autocorrelation and how a widespread statistical error has resulted in a common misconception that unskilled people are not aware of their inabilities.
Economics From the Top Down | Blair Fix

Counterfactual Evaluation for Recommendation Systems

Are you training your Recommendation System to solve "observational" or "interventional" problems? In this post, Eugene Yan examines the difference between the two and shows how counterfactual evaluation can address issues with typical observational approaches. 
Eugene Yan

Bayesian Rock Climbing Rankings

You might think that Bayes doesn't have much to do with rock climbing, but this article by Ethan Rosenthal might just change your prior beliefs. This is a great step-by-step tutorial that introduces Bayesian modeling in Stan using rock climbing data.
Ethan Rosenthal

Superb AI : Automate Data Preparation

Preparing high quality vision datasets is a long and cumbersome process. This is why companies are now using automation to accelerate labeling and QA. Take a dive into automating data preparation with Superb AI and whether or not automation is right for you and your team. 
// sponsored

Resources

Improving Your Statistical Inferences

This new book will help you improve statistical inferences, design better experiments, and report scientific research more transparently. Using helpful visuals and inspiration from the author's previous work, the book covers topics such as p-values, effect sizes, bias detection, reproducibility, and lots more. Free to read online.
Daniël Lakens

Career

Comparing Quant UX Researchers w/ Data Scientists

If you're good with data but like working with people more than models, Quantitative UX Research might be a good fit for you. Here's a great overview of what the role entails, how it compares to other data roles, how to get into it, and lots more.
Counting Stuff | Randy Au

How did you like this issue of Data Elixir?

👎    1     2     3     4     5   👍

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