Evaluating Models, Biases, and Errors

Presenter: Benedict Kuester and Jane Zanzig (Center for Data Science and Public Policy)

Level:  All

When building predictive systems, we need to estimate how well our models will perform, characterize which entities they propose for intervention, and check if they introduce unintentional bias. In this workshop, we will take an applied look at these problems, and introduce descriptive statistics that we commonly use in our projects. Basic knowledge of Python and some familiarity with data analytics will be of advantage. Participants should bring their laptops with Python3 installed.

Audience : All levels are welcome, but some prior experience working with data and familiarity with predictive modelings in the machine learning framework will be useful.