ISSUE 354  ·   September 21, 2021

Insight

2021 Data/AI Salary Survey

O'Reilly's latest salary report explores survey responses from over 3000 data professionals about pay, job stability, industries served, and career development. Includes insights about salary trends and key factors, such as education, certifications, specific skills, and lots more.
O’Reilly Media

Participatory data stewardship

These days, data breaches, scandals, and high-profile mis-uses of personal data seem rampant and ultimately undermine people's trust in data-powered initiatives. In this new report, the Ada Lovelace Institute proposes a framework that would help re-establish trust and empower people to help inform, shape and even govern their own data.
Ada Lovelace Institute

Sponsored Link

Data + Sales: Bring product metrics to the sales team

Sales leaders need accurate data to meet their quarterly targets. Use this guide to learn how to weave product metrics into Salesforce with Mode so they can accelerate the sales cycle. Download guide.

Tutorials, Projects & Opinions

How percentile approximation works (and why it's more useful than averages)

Nice primer on percentile approximations. This post breaks down how percentiles are different than averages and shows the dangers of using averages for large time-series datasets. Includes a discussion of how percentiles work in PostgreSQL, the problems with that approach, and how percentile approximation is supported in TimescaleDB.
Timescale | David Kohn

Erik Bernhardsson - Former CTO @ Better.com

There's an awesome interview series on Eugene Yan's new site, ApplyingML, that focus on practical insights from the trenches. This interview with Erik Bernhardsson is a good one to start with. It's a great read with insights about ML system development, building teams, and Erik's career path in data. Follow the "Mentors" link at the top for more.
ApplyingML

Exploring 120 years of timezones

Timezones are a pain and that's especially true for anyone who works dates in their data. This post explores why timezones are so complex and how, over 120 years, 284 timezones evolved to what we have today.
Scott Logic | Colin Eberhardt

Sourcing Training Data with Mechanical Turk

Sourcing training data using Mechanical Turk isn't new, but the process described here is a clever approach that uses gamification and different types of tasks to create a more effective data collection pipeline.
New York Times R&D

Ace your next data science interview

Get better at data science interviews and sharpen your skills by solving a few questions per week. Join thousands of other data scientists and analysts practicing for interviews! Subscribe to our newsletter.
// sponsored

Code & Tools

Awesome Data Labeling

Great collection of open-source labeling tools covering a variety of data types, including text, images, audio, time-series, and video.
GitHub | Heartex Labs

Business Intelligence for Modern Data Teams

Most BI tools use a drag and drop interface, which works for non-technical users but more and more, analysts want tools that enable common engineering practices, such as version control, testing, and abstraction. Evidence is a new open-source tool that uses SQL and Markdown to do all that and more.
Evidence

Data Visualization

Horizon Plots for ggplot2

Horizon plots are compact versions of line charts that are particularly useful for comparing time series data across different groups. This is a great post that shows how horizon plots work and introduces a new package that makes them easy to create with ggplot2. Includes lots of code snippets and screenshots along the way.
ggHoriPlot | Iker Rivas-González

How did you like this issue of Data Elixir?

👎    1     2     3     4     5   👍

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