Eva Sasson runs product growth at San-Francisco based startup Persona, with a background and interest in data science and ethics, and previous work experience at Sentry, Twilio, and Google. Sasson holds an MSc in Business Analytics and Management Science from University College London, where she explored building data science models in Python. Sasson presented about Network Graphs at the Sunbelt Conference in Utrecht, Netherlands, Pycon Canada, PyBay and about Machine Learning Bias at DataDay Mexico, in addition to speaking engagements at the United Nations Human Rights Counsel in Geneva. Sasson’s passion is to support women and underrepresented communities in tech, in addition to transitioning to a zero waste lifestyle and keeping lots of things in jars.
How Do Algorithms Become Biased?
In this talk, we will walk through the steps of how to build an algorithm to predict property prices from a dataset of property listings, focusing predominantly on finding the right features to include in building the model. Then, we will understand where in the feature engineering process we start introducing bias into our algorithms, and what are the ramifications of this if the model were to be deployed in the real world. Using the prediction algorithm as a framework to look at each step of the building process, we will also look at real-world examples of when certain decisions have led to unequal and biased results.