ISSUE 361  ·   November 9, 2021

In the News

What Data Folk Were Saying about Zillow

The data community has had a lot to say about Zillow's decision to shut down its instant-buying service. A quarter of the data science team is being laid off and there's a lot of finger pointing about why the service failed. This post by Randy Au isn't a "what happened" post. Rather, he surveys the discussions and pulls out themes and highlights. This is a fantastic overview of the situation with links to key posts along the way.
Counting Stuff | Randy Au

Avoiding Data Disasters

Rachel Thomas explores how things can go disastrously wrong on data science projects when data work is undervalued, data is used in contexts that it wasn’t gathered for, or human factors in the data science pipeline are ignored.
fast.ai | Rachel Thomas

Sponsored Link

Accelerate business insights with third-party data

Learn how companies consume financial exchange data natively through AWS Data Exchange, making consumption and analysis more efficient, agile, and accessible.

Tutorials, Projects & Opinions

A non-technical guide to interpreting SHAP analyses

Interpretability is becoming increasingly important for machine learning projects but techniques such as SHAP analysis produce complex outputs that are hard for non-experts to understand. This post is a great how-to guide for explaining SHAP results to non-technical stakeholders and is also a useful introduction to SHAP for non-experts.
Aidan Cooper

The programmer's brain in the lands of exploration and production

In her latest post, Vicki Boykis explores two distinct modes of data work and how the tools we use naturally tend to support one or the other. 
Understanding the nuances between the two modes will help you think through programming tasks and get the most from your tools.
Vicki Boykis

A guide to modeling proportions with Bayesian beta and zero-inflated beta regression models

Everything you ever wanted to know about beta regression! Use R and brms to correctly model proportion data, and learn all about the beta distribution along the way.
Andrew Heiss

A method for measuring analytical work

If the point of analysis is to help people make decisions, then there's really only one way for analysts to define success...
Benn Stancil

Announcing Innodata’s New SaaS Data Annotation Platform 

This new web-based, annotation platform reduces the cost of AI/ML projects while enabling users to develop more accurate models. With easy-to-use workflows, customizable workbenches, real-time KPIs, and auto-annotation capabilities your team can create high-quality training data on-prem or on the cloud. Sign up for a free 30-day trial today! 
// sponsored

Resources

Apache Arrow Cookbook

Apache Arrow is a cross-language development platform for in-memory analytics. This cookbook provides a variety of recipes that show how to use Arrow for things like data manipulation, reading & writing data, data transfer using Flight, and more. Different versions of the cookbook cover Python, R and C++.
Apache Arrow

Data Visualization

Coloring your ggplot2 art like a pro

ggplot2 is a popular tool for generative art but it was not originally developed for that purpose and it makes strong assumptions about how color should be used. This tutorial offers strategies for getting around ggplot2's assumptions and applying colors more freely.
Claus O. Wilke

Should you stop using box plots?

After having explained how to read box plots to thousands of workshop participants, Nick Desbarats has concluded that for most people, this common chart type is unintuitive and easy to misinterpret. In this post, Nick shows why and explores alternative chart types that are easier to read and can be more informative.
Nightingale | Nick Desbarats

How did you like this issue of Data Elixir?

😡      😐      😍

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