Blog: Data Science Insights

Automatic MCMC hyperparameter sensitivity measurements in Stan -- A worked example

We will now follow up our previous post on automatic MCMC hyperparameter sensitivity with a detailed use case taken from the Stan examples. All the code to produce the results below, as well as more examples, can be found in the examples folder on the rgiordan/StanSensitivity git repo. This simple, real-life example has a number of interesting features:

Automatic MCMC hyperparameter sensitivity measurements in Stan

A Bayesian approach to statistical modeling comes with many advantages. For example, it's the only logically coherent way to model uncertainty of parameter estimates! Being Bayesian has never been easier than it is now, thanks to high-quality, easy-to-use automatic tools like Stan.

Binder 2.0 has arrived!

The Berkeley Institute for Data Science (BIDS) is pleased to announce the release on Binder 2.0! Binder is a collection of technology that makes it easy to create sharable, interactive computing environments that run in the cloud. It makes it possible to run a server that enables users to specify their computational environment with commonly-used text files (such as requirements.txt for python users).

Most developers think we should spend more time on documentation

Farewell and Thanks to BIDS

As today marks my last day in the role of Executive Director for the Berkeley Institute for Data Science (BIDS), I’d like to take the opportunity to reflect on BIDS’ journey so far and to offer my thanks to the countless people who have worked so hard, and who continue to advance a data science approach to research. It has been an honor and a privilege to work with such an incredible community, and to help establish BIDS as an active, thriving - and growing! - research organization here on the UC Berkeley campus. 

Graph Clustering Algorithms

Introducing TextThresher 1.0 beta

After 2 years of building TextThresher, I am very pleased to announce that... A demo is provided below. But first, let me tell you a bit about what TextThresher does, how people use it, and how you can get your hands on it. 

Tales from the Docathon: How to Get Communities to Write Documentation

by Chris Holdgraf and Nelle Varoquaux

Beauty vs. Function: Not a Problem in Superheat

by Kasia Metkowski Data meets narrative in Rebecca Barter’s Superheat, an R package that creates colorful and customizable heatmaps. 

Searchable Datasets in Python: Images across Domains, Experiments, Algorithms, and Learning