ISSUE 377 · March 8, 2022
Trends
MLOps Is a Mess But That's to be Expected
With hundreds of tools and confused standards and practices, MLOps today is a mess. But it's early days for machine learning adoption and as the transformation continues over the coming years, expect the dust to settle while ML-driven value becomes more widespread. Here's an insightful look at how that may play out.
Mihail Eric
Sponsored Link
Build spreadsheet features in minutes, not months
API Spreadsheets is a suite of developer tools that lets you build complex spreadsheet capabilities in your applications. Build a spreadsheet importer, save data to Google Sheets, set up auto reporting for your team, and much more. Try out the tools for free at www.apispreadsheets.com
Tutorials, Projects & Opinions
Machine learning for color design
Huemint is a web application that uses machine learning to generate colors for graphic design. This is a fantastic post that explores why machine learning is a good fit for the use-case and how it works. Read the post and play with the app. This is clearly a labor of love.
Huemint | Jack Qiao
Work like an analyst
Benn Stancil's take on the recent discussions about whether the data stack is becoming "bundled" or "unbundled" is characteristically both comical and insightful. What's happening in the data stack is more like the creation of a big stew and "where we're going, we still need roads."
Benn Stancil
How the AI Startup Experience Has Evolved
Richard Socher discusses advances in NLP, and how the experience for AI startups has changed since the early days of deep learning.
a16z future | Richard Socher, Derrick Harris
Deploy a Data Stack 3x Faster w/o Data Eng
Modern Treasury did it — see how your team can do the same
// sponsored
Code & Tools
Notebooker
Notebooker is a web application which can execute and parametrize Jupyter Notebooks as soon as they've been committed to git. The results are searchable via the web interface, essentially turning your notebook into a production-style web-based report in a few clicks.
GitHub | Man Group
Resources
Handbook of Graphs & Networks in People Analytics
In this second book in a series, Keith McNulty explores graphs and network analysis, especially as they relate to people analytics. Few people really know how to do network analysis. This book aims to change that. Free to read online.
Keith McNulty
Algorithms for Decision Making
This new book provides a broad introduction to algorithms for decision making under uncertainty. It covers a wide variety of topics related to decision making and introduces the underlying mathematical problems and the algorithms for solving them. It's not an introductory book but the beginning sections offer a nice overview. Free to download.
MIT Press
Techmeme: Essential tech news of the moment.
Techmeme is an up-to-the-minute executive summary of trending news and commentary for the tech industry. This is my go-to site for keeping up with tech trends and breaking news. Read it online or subscribe to their free daily newsletter for the the best stories from the past day. data science newsletter in your Inbox >>
Data Visualization
Stop aggregating away the signal in your data
Aggregation is standard practice for analyzing time series data but when data is simplified through aggregation, it's easy to lose the signal and the context you need for making sense of the data. This is a great deep dive that shows the issues and offers effective strategies for different use-cases. TLDR: Embrace complexity!
The Overflow | Zan Armstrong
Plot Overview for Matplotlib Users
This tutorial shows how to make common Matplotlib charts with Observable Plot. Covers a wide variety of basic and advanced plots, including interactive histograms, scatterplots, color maps, and more.
Observable | Ayodele Odubela
 |