ISSUE 392  ·   June 21, 2022

Trends

Research highlights: Meta's Core Data Science team 

Meta's research interests have a huge impact on the tools, applications, and methods that are used in the data science community. This post explores key areas of research in things like experimentation, privacy, economic modeling, statistics, and much more. This is well-organized and includes links to papers, posts and projects along the way.
Meta Research (FKA Facebook) 

Organizations

More effective data teams using the JTBD framework

As a Data Strategist at Amplify Partners, Emilie Schario has worked with dozens of data teams and has seen a consistent pattern of JTBDs (Jobs-To-Be-Done) that data teams serve within organizations. This is a great post about the JTBD framework, how it can be used to keep your team focused, and the five specific JTBDs that data teams serve.
Locally Optimistic | Emilie Schario

A Better Way to Put Your Data to Work

This article explores how successful companies are getting more value from their data by managing it like a product.
Harvard Business Review

Sponsored Link

6 best practices to save time when creating AI data

78% of machine learning projects stall before deployment, making speed a primary concern for AI teams. However, wading through unstructured data to accurately annotate assets requires a tremendous amount of patience, organization, and time. This webinar will cover six key time-saving practices implemented by leading AI teams when handling training data. Save your spot today.

Tutorials, Projects & Opinions

A framework for document processing solutions

Extracting data from PDFs and scanned documents is more complicated than you might think. In this post, Lj Miranda explores the issues and walks through a framework for designing a document processing solution. Includes links to papers, recipes, and repos along the way.
Lj Miranda

Graph Machine Learning at Airbnb

Real-world machine learning problems can often be framed as graph problems. For instance, on online platforms, users may interact and share photos, messages, etc with each other and those connections form edges that can be used to create a graph. In this post, Devin Soni introduces graph neural networks with practical applications at Airbnb.
The Airbnb Tech Blog | Devin Soni

Introduction to Graph Neural Nets with JAX/jraph

To go further with graph neural networks, here's a more in-depth introduction from DeepMind. This is easy to follow and is set up as a notebook so you can run the tutorial and interact with it.
Colab | DeepMind

Mona: Intelligent Monitoring Solution for AI / ML

Trusted by data science teams from top companies around the world, Mona provides you with complete visibility into the performance of your AI-driven products, automatically surfacing issues within your data and models. Track custom metrics for any AI use case, any industry and any tech stack. Try it for free or book a demo today! 
// sponsored

Resources

R Workflow for Reproducible Research

Great best-practices resource for reproducible research. Covers data documentation and manipulation, statistical analysis, graphics, reporting and more. It's written from the perspective of biomedical research but the methods here are widely applicable.
Frank E Harrell Jr

Data Visualization

Creating flowcharts with {ggplot2}

Flowcharts can be a useful way to visualise complex processes, especially when they can be created programmatically. This tutorial shows how to create highly customisable, good looking flowcharts using {igraph} and {ggplot2}.
Nicola Rennie

How did you like this issue of Data Elixir?

👎    1     2     3     4     5   👍

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