Nimble Machine Learning with Data Pipelines
Boston Data Science March 2019 Meetup
Abstract:
It's an open secret amongst data scientists that, despite all of the Deep Neural Network talk, we spend a surprisingly small portion of our time fiddling with models. Model improvements also often require data pipeline improvements to implement, yet there are hundreds of resources on TensorFlow and very few on how to develop a training pipeline. This talk aims to teach you everything that Matt wishes he'd known when he started, which is a lot.
Speaker Bio:
Matt Ritter has worked with many models in his 6+ years at athenahealth, but he's learned (the hard way!) that a detailed understanding of your data pipeline is the difference between a successful data science project and a world of frustration. He leads athena's Data Methods Lab series and has started a Newton/Waltham/Watertown-area "Quantified Self" group that you're totally welcome to join. Ask him anything on LinkedIn or at @mr_itter
Slides:
https://drive.google.com/file/d/1C2o7...
https://www.meetup.com/Boston-Data-Sc...