In this short write-up, I explore data on donations to a current US senator. The dataset comes from publicly available FEC receipts data (see https://www.fec.gov/data/receipts/). I first walk through an exploratory analysis to document salient donor characteristics, as well as trends in donation over time and across election types. Then, after cleaning the data and engineering relevant features, I employ a more formal machine-learning design to train an optimally-tuned model to predict what kinds of donors are likely to donate more. The exploratory analysis will turn out to be more informative, largely because the data quality greatly constrains the ultimate quality of any predictive model.
isaacrabbani / senatordonations Goto Github PK
View Code? Open in Web Editor NEWA short practice project in which I conducted an exploratory data analysis of donations to a US Senator, and then employed machine-learning tools to predict what kinds of donors are likely to donate more.