Angela Zhou, 2016, Operations Research and Financial Engineering

Traditional fisheries management policies remain uninformed by an understanding of the social dynamics and connectivities of fishing communities. Fishermen make decisions—which are affected by their social communities, different norms and economic incentives—to compete or collaborate for resources. These individual decisions play out through communication and information-sharing about fishing locations, and affect fisheries management as a whole. A competitive fishing strategy, for example, has different implications for sustainability than a more collaborative one. Unfortunately, it’s difficult to precisely measure and characterize these information-sharing behaviors, beyond reviewing anecdotal interview evidence from fishermen. GPS-reported tracks of fishermen’s movements, however, provide strong signals about fishermen’s behavior, and we can infer information-sharing signals from unexpected correspondences in this movement data. During this internship, I applied machine learning techniques to the movement data to infer the fishing behavior of the vessels, and simultaneously worked with an agent-based model of fishermen to investigate how information sharing impacted catch rates. I learned through this hands-on experience how data mining techniques can be used to investigate the connections among people and inform more sustainable management practices. This work on connecting decision theory to statistical inference from spatiotemporal data has furthered my interest in decision theory in general, connecting my interests in statistics, data science, economics, and the social sciences.