DatA Related Projects

With survey data from a 2022 book Fans Have More Friends, we sought to predict fan activities based on self-reported fandom and various demographic data towards developing more targeted marketing strategies. Using classification algorithms such as Random Forest classifiers and AdaBoost classifiers, we predicted activities such as when a person purchases multi-game tickets with an average overall accuracy of about 73%. [SLIDES] [GIT REPO] 

The NFL, a multi-billion dollar industry, reflects similar corporate challenges around hiring and diversity/inclusion. Using centrality measures such as eigenvector centrality and community detection such as k-core decomposition, I analyzed the NFL coaching network, revealing patterns of entrenched connections among coaches. This social network analysis, often used in customer churn prediction, highlights potential barriers to diversity within NFL leadership.

Zero forcing in a network is a process similar to the spread of information in a network, where an initial set of nodes spreads information through the network according to specific rules and restrictions. I have studied zero forcing and its variations, including a version that incorporates "leaks" in the network, inspired by monitoring an electrical network and guarding against faulty equipment. Python code I developed has been instrumental in generating conjectures and insights in this area.

I studied a pursuit-evasion game on networks, where one or more "cops" try to capture a "robber" moving across the network’s vertices in a turn-based manner. This research provides insights into network properties, such as the minimum number of cops needed to guarantee capture, which informs efficient resource allocation. I developed Python code to generate conjectures and derive insights in this area.