B.A. Data Science and Psychology, University of Rhode Island. Research interests include the impact of social media on mental health and the prediction of voting propensity. I use my writing to advocate for a kinder, more walkable world, specifically commenting on housing, migration, and gender policy. Op-eds featured in The Guardian, Al Jazeera, The Providence Journal, and Público.
This project aims to provide a supervised machine learning model that can take historical voter data as an input and output the probability that an individual will vote in the upcoming election. Along with this, the model should be able to predict the most probable party designation of unaffiliated voters prior to polling. These metrics will, in theory, help campaign managers to more efficiently direct their efforts towards likely voters. Our experiments successfully replicated findings in prior political science and statistical research showing that prior vote history is predictive of voting behavior. In addition, this research was able to extend prior findings to predicting voting behavior in primary elections, where the vast majority of voters do not participate in. In experiment 2, we successfully applied an artificial neural network to predict party affiliation in a partisan primary, achieving remarkable results with nearly a perfect model. Even with an imbalanced dataset, prior vote history with basic demographic variables provided reasonable results. Due to limitations in our study design, our findings represent a conservative attempt at predicting future voter behavior.
