ISSUE 359  ·   October 26, 2021

Trends

The future of operational analytics

Applications that help consumers make decisions use data behind the scenes but the data, itself, is mostly hidden. On the other hand, dashboards and BI tools are awash in data but the tools are generally divorced from the decisions they support. In this post, Benn Stancil makes a great case for a better way to make decisions that's already here — just not in the tools you use at work.
Benn Stancil

Sponsored Link

eBook: Practical Guide to Third-Party Data

To help navigate a proliferating data landscape, AWS Data Exchange has created A Practical Guide to Using Third-Party Data in The Cloud. Implementing the best practices in this collection will help provide leaders with foundational knowledge and actionable advice to create value, drive innovation, and transform your business with third-party data. Download Now

Tutorials, Projects & Opinions

Interpreting A/B test results: false negatives & power

This introduction to false negatives and statistical power is the fourth post of a multi-part series on how Netflix uses A/B tests to inform decisions. This is turning into a great series that uses simple thought excercises to build intuition around A/B testing concepts.
Netflix Tech Blog

How and why we built a custom gradient boosted-tree package

Lyft built their own gradient boosted tree package for working with spatial data called GeoBoost. It's slower to train than off-the-shelf packages, but can be customized to treat space and time more efficiently and yield less volatile predictions. Here's how it works.
Lyft Engineering | Or Cohen

Writing a book for O'Reilly

O'Reilly Media is the popular tech book publisher that puts animals on its covers. If you've ever thought about what it takes to write one of those books, this is a fantastic, first-hand account of what's involved and how it can be worthwhile.
Mark Edmondson

Getting Strategic Value from Data Analytics When Initial Attempts Fail

Get your copy of this free Harvard Business Review report to learn about the state of data analytics, the obstacles preventing its progress, and how organizations can produce more value. Get the report.
// sponsored

Code & Tools

Malloy

Malloy is an experimental language for describing data relationships and transformations. It will feel familiar to SQL users but it's intended to be easier to reason about than SQL and computations are modular, composable, reusable, and extendable in ways that are consistent with modern programming paradigms.
GitHub | Looker Open Source

Introducing Hamilton, a microframework for dataframe generation

Creating dataframes is easy but for large domains, it can be challenging to manage the associated codebase. Hamilton is a new open-source framework that lets you create dataframes from functions and helps to manage complex pipelines. You don't need to maintain the dataframe and you don't need glue code. Here's how that's useful.
Stitchfix | MultiThreaded

Resources

Introduction to Probability for Data Science

Data science books are often  written for either programmers — with an emphasis on how to do things — or for mathematicians — with an emphasis on theory. This new text takes a middle path and integrates classical probability theory with lots of examples in Python, Julia, R, and Matlab. Free to download.
Stanley H. Chan

Data Visualization

Superpowers as Inspiration for Visualization Design

We already talk about visualizations as amplifying cognition, but what if we want to make new visual tools that really feel like superpowers? Great post!
Multiple Views | Calgary Interactions Lab

How did you like this issue of Data Elixir?

👎    1     2     3     4     5   👍

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