Introduction to Using rubicon-ml (VIDEO)
How this open source data science tool helps capture reproducible information throughout your model’s development
By Srilatha Ranganathan, Sr. Mgr, Data Engineering; Joe Wolfe, Master Software Engineer, ML Tools; and Ryan Soley, Software Engineer, ML Tools
Recorded for PyCon 2021, this sponsor workshop video serves as a basic introduction to the rubicon-ml data science tool and how it helps the model development process. Tailored for data scientists and model developers of all levels, this video highlights the importance of tracking data to gain full reproducibility during model experimentation; as well as how rubicon-ml can help organizations fulfil this need.
The video also includes a demo exploring the rubicon-ml library itself. This demo shows exactly what rubicon-ml is, how to use it, basic best practices, and the various integrations it offers.
After watching this video introduction to rubicon-ml, we hope you’ll take away:
- The importance of offering reproducible experimentation throughout the model training process.
- How to integrate rubicon-ml into your model pipelines to automatically track (and later explore) experiment data over time.
- How to use rubicon-ml to prepare a "story" of the model development process via the dashboard.
- How to link your training data directly to the corresponding model code that produced it.
As a side note, this video introduction to rubicon-ml is tailored to the model development process. But it's worth noting that rubicon-ml can be used outside of that for any logging over time use case. For example, rubicon-ml can be used to track performance metrics of a web service over time, and as a way to capture API logs.
For more information on rubicon-ml please visit our GitHub repository.