Opening the Black Box

Talk

Go to NumFOCUS academy page.

Data scientists have likely heard that machine learning models are ‘black boxes’ and not interpretable. This notion, once formerly true, is no longer correct. In this talk, the modelers at Southeast Toyota Finance will focus on model-agnostic interpretability techniques. We’ll provide an end-to-end example of the experimentation process of modeling combined with interpretability techniques.

Speakers

Ben Fowler

Ben Fowler has been in the field of data science for over five years. In his current role at Southeast Toyota Finance, Ben leads the end to end model development process with Chelsey Meise to solve problems of interest.

This process involves: • Discussing and defining the problems of interest for the business • Designing and implementing a plan for sampling of the data and conducting experiments • Exploratory data analysis • Data cleaning and preparation • Feature engineering and feature selection • Model evaluation and selection • Model interpretability • Presenting results • Production code for reproducibility and serving of models in deployment on new data • Model documentation • Researching novel technologies and machine learning methodologies

Ben holds a Master of Science in Data Science from Southern Methodist University, graduating in 2017. Following graduation Ben has been a guest speaker to the SMU program multiple times. Additionally, Ben has spoken at the PyData Miami 2019 and PyData Los Angeles 2019 Conferences along at Predictive Analytics World Finance 2020. Furthermore, Ben was selected to be a speaker at the O’Reilly Strata Data and AI Conference in San Jose in 2020 and appeared on the Datacast podcast.

Chelsey Kate Meise

Chelsey has a background in statistics, having spent the last ten years working as a predictive modeler for various companies ranging from healthcare to finance. She graduated from Villanova University with a Master of Science in Analytics and studied Applied Statistics and Health Science in undergraduate school at Kennesaw State University. She has nine major journal publications from her time doing research at University of Pennsylvania. In her current role at Southeast Toyota Finance, Chelsey works on the machine learning team helping to manage credit risk.