ISSUE 256   ·   October 22, 2019        

Insight

Machines Beat Humans on a Reading Test.
But Do They Understand?

Nice introduction to recent NLP advances for smart, non-experts.
Quanta Magazine

Can you make AI fairer than a judge?

The US criminal system uses algorithms to help determine who should be jailed and who should be released and it's messy. In this interactive article, Karen Hao and Jonathan Stray walk through the algorithm known as COMPAS and show how hard it is to quantify fairness.
Technology Review

Data — from objects to assets

Over the past 150 years, ideas have shifted drastically as to what counts as data, which data are reliable and who owns them. Here's how the idea of "data" has evolved over that time and gotten to be so big.
Nature

Sponsored Link

"No Bias" A Podcast About the Frontier of ML

Each episode your host Melody chats with machine learning researchers Nikhil & Saurabh about the frontier of AI & ML technology. Check out our latest episode: Bias in ML-The Good, the Bad, & the Ugly. They discuss helpful mathematical bias & cognitive bias skewing data & compromising models. Listen now.

Tools and Techniques

SQL query steps

"When this SQL query runs, here's how I think of what happens... " Great thread!
@b0rk ( 🔎Julia Evans🔍 )

Time Series Prediction
A short introduction for pragmatists

In this introduction to working with time-series data, Thomas Ebermann compares prominent techniques and shows how to implement them. This is a well-organized post that's easy to follow with plots and code snippets along the way. Covers Winter-Holt models, SARIMA, Neural Networks and Facebook's Prophet library.
L//P

Predicting and monitoring payment volumes with Spark and ElasticSearch

In this post, Andreu Mora describes the design considerations and implementation decisions for a machine learning system that scores in real-time and is deployed at scale. This is a real-world design that's used by the Adyen Payments platform.
Adyen Blog

Introducing Comet: Github for Machine Learning

Join thousands of data scientists who use Comet.ml.  Build reliable ML models for real-world applications.  Automatically track, compare, explain and collaborate around your ML models and experiments.  100% Free for public projects! Sign up now.
// sponsored

Career

Updating the Stack Overflow Salary Calculator

Stack Overflow's 2019 Salary Calculator supports a wide variety of regions and technical job titles around the world. In this post on the Stack Overflow blog, Julia Silge offers a nice introduction to the data with insights into how the model works. Check out the post first and then head over to the calculator to see how your salary compares.
Stack Overflow

Machine learning portfolio tips

Great tips and insights here for figuring out good projects to include in a portfolio of ML work. Includes specific resources for generating ideas, things to watch out for, and practical ways to attract interest. Also, be sure to check out the suggested readings in the final bullet.
@EmilWallner

Data Viz

How much “fake news” can we identify on Twitter?

This is awesome data viz storytelling.
Knight Foundation

Conferences and Events

Scale by the Bay - Learn from top Data Engineers from Netflix, Spotify, Twitter, DataStax, Databricks how to build an end to end data pipeline and seamlessly integrate deep neural networks with traditional software development. November 13-15. Data Elixir readers save 15% with this code: DATAELIXIR15
More >>

Data Elixir is curated and maintained by @lonriesberg. For additional finds from around the web, follow Data Elixir on LinkedIn, Twitter or Facebook.