Bootstrapping labels, rebranding data, visualizing ordinal variables

Rebranding data

When the “Big Data” hype crashed, it was quickly replaced by “data science.” New branding, a few new companies, but essentially the same thing minus a few inflated expectations. So what’s next?

Data quality unpacked

Andrew Ng surveyed studies that showed how improving the quality of the data significantly improves model accuracy, while improving the model algorithm had little to no effect. Here, Kenn So and Ben Lorica take a look at data quality and how to achieve it.

Explaining ML to a child...and to an expert

In this video, Hilary Mason explains machine learning to a child, a teen, a college student, a grad student, and an expert.

Bootstrapping labels

Most machine learning tutorials assume the availability of training labels. But in real life, labeled datasets aren’t always available. What then?

IN COLLABORATION WITH Algorithmia

Free report: The Framework for ML Governance

Effective governance is a crucial component of every successful machine learning initiative. And while there’s lots of talk about governance, very little prescriptive advice exists for how to implement it. Until now.

And it’s free, courtesy of Algorithmia.

Visualizing ordinal variables

They’re not numeric. They’re not categorical. The most common way of analyzing ordinal data is to just assume it’s metric. Octavio Medina explains why that approach can fail you, and how to account for the effect of a continuous variable on the ordinal outcome.

How to evaluate the quality of a data pipeline

What questions should you be asking yourself when evaluating a data pipeline? Start with these.

+ Build a Robust Data Pipeline with Airflow, dbt, and Great Expectations, a live online O’Reilly course, Sept. 14

IN COLLABORATION WITH Acquia

Free ebook: Transforming Customer Data into Insights

Transform your marketing programs by tapping into the power of customer data with this handy guide. You’ll get case studies that show how successful organizations have created a customer-first experience, and a look at the challenges marketers face—and the technology solutions that have risen as a result, including customer data platforms.

And it’s free, courtesy of Acquia.

COVID-driven cloud surge takes a toll on the data

COVID-19 accelerated cloud migration at many companies—it was a matter of survival. As Microsoft CEO Satya Nadella said, “We’ve seen two years’ worth of digital transformation in two months.” But in many companies, data management was an afterthought. Here’s a look at some of the data issues that arose.

+ GCP Certification Prep Crash Course: Professional Data Engineer, a live online O’Reilly course, Sept. 30

How AI Will Change Our World by 2041 with Kai-Fu Lee and Tim O’Reilly

O’Reilly members, join us on September 9, from 4pm to 5pm PDT, for what we expect will be a fascinating conversation between Kai-Fu Lee—celebrated AI pioneer, former president of Google China, and best-selling author of AI Superpowers—and Tim O’Reilly. They’ll explore predictions from Dr. Lee’s new book, AI 2041: Ten Visions for Our Future, and discuss how current AI technologies, such as deep learning, computer vision, robotics, autonomous driving, and natural language processing, can provide insights into future advances and challenges that will have far-reaching impacts across the globe.

Not yet an O’Reilly member? Try a 10-day free trial.

Without code for DeepMind’s protein AI, this lab wrote its own

DeepMind solved a fundamental problem in biology but didn’t share the code for seven months. So a University of Washington team tried to recreate it.