Webcast recap: Enabling data science with GitHub
Data powers much of the software we use every day. Although data science teams operate differently than engineering teams, they can apply the same best practices that engineers use to…
Data powers much of the software we use every day. Although data science teams operate differently than engineering teams, they can apply the same best practices that engineers use to share code, communicate, and work together more efficiently.
In a recent webcast, GitHub Senior Solutions Engineer Bryan Cross shows us how data scientists use GitHub to support their workflows. Below, you’ll find three key takeaways from the presentation, as well as the on-demand link.
Experimentation
With the ability to snapshot versions of your work, you can iterate and experiment freely without the worry of losing previous work. If you hit a dead end, back up to a previous version and continue from there. Even your dead ends are preserved on GitHub—giving your team a complete record of everything that did and didn’t work.
Discoverability
With tools like GitHub search, finding and sharing work is no longer a chore. Using basic functions, your team can search for files, people, repositories, and specific conversations with search terms and more for in-depth discovery. When starting a new project, easily find what work has already been done on the topic and avoid duplicating efforts.
Collaborative work with GitHub
With issues and pull requests, you can seamlessly work together or cross-functionally with other development teams. Issues act as threaded discussions that loop in team members and other stakeholders. With issues, you can discuss how best to tackle a new project and keep everyone updated on the project’s progress. And with pull requests, your team can review code, results, and commentary, suggesting ways to improve all three before sharing with stakeholders.
Sharing results
Once your work is written, reviewed, and ready to go, you’ll want to share your results. GitHub renders Jupyter Notebooks hosted in GitHub repositories, making notebook sharing a breeze. If you prefer, generate a markdown document from your RMarkdown Notebook and GitHub will render it automatically. To share your results as a webpage, GitHub Pages provides an easy way to host a simple website. Best of all, these are all hosted in a repository, so results benefit from all the snapshotting, search, and collaboration tools discussed above.
To learn more about incorporating GitHub into your data science workflows, watch the webcast. You can also register for upcoming events or watch previous webcasts on our Resources Page.
Written by
Related posts
Enhancing the GitHub Copilot ecosystem with Copilot Extensions, now in public beta
Whether you’re an individual developer looking to streamline your workflow or an organization aiming to integrate proprietary tools, GitHub Copilot Extensions now offers a platform to make that happen and to share your creations on the GitHub Marketplace.
First Look: Exploring OpenAI o1 in GitHub Copilot
We’ve tested integrating OpenAI o1-preview with GitHub Copilot. Here’s a first look at where we think it can add value to your day to day.
GitHub Availability Report: August 2024
In August, we experienced one incident that resulted in degraded performance across GitHub services.