ISSUE 325 ·   March 2, 2021        

Insight

A View Of The Future Of Our Data

What if instead of giving our data away, there were "data coalitions" that acted as brokers and made sure that data was used on behalf of its members’ interests? For now, it's an optimistic idea but it could become real in the not too distant future. Here's how that could look.
Noema Magazine | Matt Prewitt

Can Auditing Eliminate Bias from Algorithms?

Algorithmic auditing is an emerging industry that examines black box algorithms to ensure that they're free from bias. Given the persistent issues around bias and the high stakes involved, you might think that auditing would be a welcome solution. Not exactly.
The Markup

Gartner Top 10 Data and Analytics Trends for 2021

From artificial intelligence to small data and graph technology, here are 10 key trends that data and analytics leaders should consider for 2021.
Gartner

Sponsored Link

This Asset Class is Returning 13.6% Annually

84% of ultra-high-net worth individuals collect art, according to a 2019 Deloitte survey. It makes sense—contemporary art returned 13.6% per year over the last 25 years vs. 8.9% for the S&P 500. Masterworks let you invest in million-dollar works by artists like Basquiat and Banksy at a fraction of the cost. Data Elixir Subscribers can sign up and skip the 17,500 person waitlist today.*

*See important info

Tutorials, Projects & Opinions

Visualizing Data Timeliness at Airbnb

In a large organization like Airbnb, data that isn't available when it's expected to be available can adversely affect multiple teams and products. As part of its series on data quality, this post walks through the issues of data timeliness at Airbnb and the considerations for an internal tool that tracks timeliness and lineage across pipelines. This is a great post with lots of  screenshots from their "SLA Tracker" tool.
Airbnb Engineering & Data Science

Feature Stores - A Hierarchy of Needs

Eugene Yan's latest post is a fantastic introduction to feature stores. It's organized as a "hierarchy of needs," which makes it easy to understand what feature stores provide exactly, and which capabilities may be useful for your particular needs. Includes lots of examples.
Eugene Yan

Is Facebook's "Prophet" the Time-Series Messiah?

Facebook prophet is a popular package for time series prediction and has been downloaded nearly 14 million times.  But does it work?
Microprediction | Peter Cotton

How to break a model in 20 days

What can go wrong with a machine learning model in production? In this tutorial, learn how to train a model, simulate deployment, and analyze its gradual decay.
Evidently AI

Map of my personal data infrastructure

This "data liberation infrastructure" map is for the seriously self-quantified. It's part crazy, part geeky, and part awesome. The write-up includes links to posts that elaborate on different parts of the map, along with the scripts and tools that are used to maintain it.
beepb00p

Sign up to get Data Elixir's  data science newsletter in your Inbox >>

Data Elixir is curated and maintained by Lon Riesberg. If you have questions or suggestions for the newsletter, just reply back to this email.