Improving Online Shopping Experience Through Data Collection; Machine Learning Bias Within the US Justice System

Mirkhah, Bijan, School of Engineering and Applied Science, University of Virginia
Vrugtman, Rosanne, EN-Comp Science Dept, University of Virginia
Ferguson, Sean, EN-Engineering and Society, University of Virginia

Data collection and analysis is one of the most vital ways that companies can gather insights on their customers. When done properly, the insights gathered from the user can improve the users experience by providing refined results, and improve the companies by allowing for better advertisement of their products. In the big data economy that we live in, advertising and marketing only represents a fraction of what data is used for. The following technical and STS discuss data collection and analysis. While the technical demonstrates how data should be collected and analyzed, the STS thesis expands on the potential for data misuse. The technical thesis is a summary of my internship over the summer with a online retail company. Thought the internship, the end users of the data and the people that the data impacted were the first considerations when designing the system, and the project was built from the end customers in mind. This allowed for the system to have the greatest benefit for the customers that the data impacted, as well as benefit the other teams that would be using the data for insights. In the end, the project that I created allowed for collection of data and will be used by future teams for their own analysis and insights. The STS thesis explores the pitfalls of data analysis in the world of policing. The paper explores two case studies: COMPAS (predictive sentencing) and PredPol (location based police predictions). The algorithms were designed without the end user in mind therefore the algorithms ended up having bias and adversely impacting minority groups. Ultimately, conducting the STS research allowed for me to understand how algorithms can fail and its weaknesses, therefore I have learned how to best avoid the mistakes of the data analysts that designed them. I feel like learning about algorithmic bias has allowed for me to improve as a data engineer and a software engineer.

BS (Bachelor of Science)
machine learning, algorithmic bias, data engineering

School of Engineering and Applied Science
Bachelor of Science in Computer Science
Technical Advisor: Rosanne Vrugtman
STS Advisor: Sean Ferguson
Technical Team Members: Bijan Mirkhah

All rights reserved (no additional license for public reuse)
Issued Date: