Support the ECS Data Sciences Hack Week

ECS Data Science Hack WeekBuilding on the success of the first ECS Data Sciences Hack Day at the 232nd ECS Meeting this past October 2017, ECS is pleased to offer another opportunity at the 233rd ECS Meeting in Seattle this May.

ECS Data Sciences Hack Week is the Society’s foray into building an electrochemical data sciences and open source community from the ground up. Dataset sharing and open source software have transformed many “big science” areas such as astronomy, particle physics, synchrotron science, protein and genomic sciences, as well as computational sciences. The goal of this event is to increase awareness and impact of data science tools, open source software, and shared datasets in electrochemistry by bringing together people from different backgrounds to collaborate.

Data science tools and approaches have the potential to transform bench science like electrochemistry. The critical need is to build a community of electrochemical data scientists, the people who will contribute to a growing library of shared experimental and computational datasets, and who develop and adapt open source software tools.


By: Neal Dawson-Elli, Seong Beom Lee, Manan Pathak, Kishalay Mitra, and Venkat R. Subramanian

This article refers to a recently published open access paper in the Journal of The Electrochemical Society, “Data Science Approaches for Electrochemical Engineers: An Introduction through Surrogate Model Development for Lithium-Ion Batteries.”

Electrochemistry and Data Science

Image via Neal Dawson-Elli
(Click to enlarge.)

Data science is often hailed as the fourth paradigm of science. As the computing power available to researchers increases, data science techniques become more and more relevant to a larger group of scientists. A quick literature search for electrochemistry and data science will reveal a startling lack of analysis done on the data science side. This paper is an attempt to help introduce the topics of data science to electrochemists, as well as to analyze the power of these methods when combined with physics-based models.

At the core of the paper is the idea that one cannot be successful treating every problem as a black box and applying liberal use of data science – in other words, despite its growing popularity, it is not a panacea. The image shows the basic workflow for using data science techniques – the creation of a dataset, splitting into training-test pairs, training a model, and then evaluating the model on some task. In this case, the training data comes from many simulations of the pseudo two-dimensional lithium-ion battery model. However, in order to get the best results, one cannot simply pair the inputs and outputs and train a machine learning model on it. The inputs, or features, must be engineered to better highlight changes in your output data, and sometimes the problem needs to be totally restructured in order to be successful.


By: Matt Murbach, University of Washington

Hack Day

Co-organizer David Beck led a hack session during the ECS Data Sciences Hack Day.

The full vibrancy of the electrochemical community was on show during the recent 232nd ECS Meeting in National Harbor, MD. Adding to the diversity of ideas and excitement for electrochemistry were the 30 participants of the inaugural ECS Data Sciences Hack Day on Wednesday, October 4. The participants in the hack day traveled from around the globe and represented varying stages of careers in both academic and industry roles.

The day-long event was kicked off with a short series of informational sessions covering some of the essential tools in any data scientist’s toolbox. During lunch, participants pitched their ideas for projects, and teams for the afternoon session organically formed around common interests. The remaining time during the afternoon was reserved as open “hacking” time for working on the project ideas. Excitingly, good progress was made in this four-hour block with teams working on a wide variety of projects, including:


Open Science and ECS

On October 4, during the Society’s 232nd meeting, ECS will be hosting its first ever ECS Data Sciences Hack Day. This event will be ECS’s first foray into building an electrochemical data sciences and open source community from the ground up.

On this episode of the ECS Podcast, we discuss the upcoming ECS Data Sciences Hack Day, the importance of dataset sharing, how open source software can transform the field, and the future of open science.

This episode’s guests include Daniel Schwartz, Boeing-Sutter Professor of Chemical Engineering and Director of the Clean Energy Institute at the University of Washington; David Beck, Director of Research with the eSciences Institute at the University of Washington; and Matthew Murbach, president of the University of Washington ECS Student Chapter.

Schwartz, Beck, and Murbach will be at the 232nd ECS Meeting this fall in National Harbor, Maryland participating in OpenCon and running the ECS Hack Day. There’s still time to register for both of these events.

Listen to the podcast and download this episode and others for free on Apple Podcasts, SoundCloud, Podbean, or our RSS Feed. You can also find us on Stitcher and Acast.