[IN-PERSON] Jupyter notebooks extensions to improve ML workflows

Jupyter notebooks, weren’t designed for machine learning projects, but we can and should extend them to fill in missing functionality that’ll help us build better models faster.

In this talk, I’ll argue that the missing key features Jupyter notebooks need to improve their ML workflows are ones that software engineers already enjoy in their development environments. (These features leverage metaprogramming, automated testing, and easily hackable editors.) Next, I’ll introduce key modules and extension points in python and ipython you can use to build out these missing features. (viz., python’s ast module to manipulate cell code and ipython’s event hooks and rich display methods) Finally, I’ll briefly show how data chimp uses these low-level building blocks to make it easy to hack your notebook into an ML-first tool.

Matt is a wannabe philosophy professor turned wannabe tech entrepreneur. He's spent most of his career writing mobile and web software and most recently, he worked as a data science engineer at Heap. Now he's working on data chimp, a tool to help data scientists analyze data faster.

IMPORTANT!! The OCLS Melrose Center new guidelines enforce a maximum attendees for the LED Wall space of 15 persons. RSVP as soon as possible. If there is change of plans and cannot make it, please undo your reservation, so other members can join the meetup.