Women in Data Science @ Penn Conference
This is What a Data Scientist Looks Like
On February 12, 2021, Vice Dean of Analytics at Wharton Eric Bradlow welcomed attendees to the final day of the Women in Data Science (WiDS) at Penn Conference, which highlighted student-led projects. This video features:
A Post-Processing Algorithm for Bias Reduction in Big Data Analytics with Brian Handen, Hyewon Lee, Margaret Ji, Tashweena Heeramun, and Trishla Pokharna
Currently, due to poor data quality and lack of data for under-represented groups, we see several populations being negatively impacted in fields ranging from healthcare to criminal justice. To target this problem, our project aims to promote subgroup fairness in black-box algorithms that make important decisions by creating a post-processing algorithm. In order to do so, we are implementing the multicalibration algorithm which enforces stricter bounds on what it means for a predictive model to be both accurate and fair by reducing the implicit bias reported in existing data sets.
About WiDS @ Penn Conference
Women in Data Science (WiDS) @ Penn is an independent virtual event presented by the University of Pennsylvania to coincide with the annual Global WiDS Conference held at Stanford University and an estimated 150+ organizations worldwide. All genders are invited to attend WiDS regional events, which features outstanding women doing outstanding work. This year’s theme for WiDS Philadelphia @ Penn was: This is What a Data Scientist Looks Like.