About us#
Hi there, we are the FACET team! We are part of BCG GAMMA, a global team of data scientists working across all regions and industries who love what we do.
Our Story#
Industry standard packages for machine learning (scikit-learn) and model inspection (SHAP) are critical for best practice data science. However, combining these into an efficient and reproducible process that yields deep insights for industrial and business processes can be challenging and time consuming. We found this to be a common challenge for our global data science teams across 1000s of projects over the years.
So what did we envision to solve this problem? Well, an easy way to create a robust pipeline for learner selection with cross-validation, advanced model inspection to deeply understand feature contributions to predictions, and the ability to use simulation to gain insight into optimal feature values. At the same time we wanted to build from the industry standard eco-system (i.e., scikit-learn, SHAP) not re-invent it.
So we got a bunch of data scientists to look at the problem from every angle and, well, you get FACET! Little geometry humor there, we apologize. FACET implements a newly developed algorithm to explain how features of a predictive model collaborate to predict outcomes by identifying patterns across the explanations of many individual predictions. This global perspective complements the local perspective of SHAP. Further, this advanced model inspection allows strong independent features to be identified and then explored via simulation to understand optimal values for the predictive target of interest.
Our Team#
Of course, open-sourcing FACET would not have been possible without the support, contributions and perspectives of many fantastic people! FACET, many perspectives, okay that was the last geometry pun, we promise.
FACET Team at BCG GAMMA
|
|
|
|
|
|
|
|
|
|
Jason Bentley
Malo Grisard
Konstantin Hemker
Jan Ittner
Sithan Kanna
Ricardo Kennedy
Florent Martin
Jörg Schneider
Andy Shora
Mateusz Sokół