Enhancing the reproducibility of computational notebooks in ecology and environmental sciences
This workshop is aimed at familiarizing attendees with some of the many ways in which the reproducibility of computational notebooks can be compromised, and how this affects the validity of associated scientific results. While the focus here will be on computational notebooks run in Python or R via Jupyter and RStudio, the majority of the highlighted issues apply to other notebook formats and programming languages as well.
Benefits to participants: Participants of the workshop will be sensitized to the different ways in which computational reproducibility may be compromised and what precautionary measures they can take to enhance the reproducibility of their own research. They will also gain insights into how to address reproducibility issues when attempting to reproduce research originating from others.
Skills to be imparted: Participants will be familiarized with the recommended practices for sharing computational notebooks, particularly via Jupyter and R. They will learn how to structure, share, review, and rerun a computational notebook such that reproducibility of their own research is maximized
Targeted audience : This workshop is intended for actual and would-be users of computational notebooks (via Jupyter or RStudio) with at least a basic familiarity with Python or R
Format of the Workshop: After an introductory overview of how computational notebooks can facilitate reproducibility, a number of common problems will be highlighted. On that basis, strategies to maximize computational reproducibility of such notebooks will be presented based on practical examples. These examples will be drawn from two main sources: (i) until a week before the event, workshop attendees will be invited to submit computational notebooks of their own or from published studies, (ii) a systematic study of computational notebooks associated with publications.
Participants are then invited to work in groups on addressing reproducibility issues found in the example notebooks, and to apply the presented strategies to overcome such issues. This group work will be facilitated by the workshop organizers.
● Daniel Mietchen, School of Data Science, University of Virginia
● Sheeba Samuel, Friedrich Schiller University Jena, Germany
● Luiz M. R. Gadelha Jr., Friedrich Schiller University Jena, Germany
● Aaron Willcox, University of Melbourne, Australia
● Elliot Gould, University of Melbourne, Australia