While a great deal of my work at Pew has been focused on improving my team’s workflows, educating staff on how to use data science tools, and quality checking, I have also contributed to a number of reports, primarily focusing on Twitter use by U.S. adults.
Many Chilean workers face significant workplace safety issues. Dirección del Trabajo, Chile’s labour agency, is responsible for reducing the risks workers face through performing inspections and remediations. However, a lot of time has been spent inspecting facilities reactively based on complaints. My team built a pipeline in Python to move data from its raw format to clean Postgres databases, run machine learning models, and evaluate the results of these models to help DT identify facilities to prioritize in future inspections. This was complicated, and messy, so feel free to reach out for more details.
As a volunteer project manager with DSSG Solve, I led a team working with ILAO to find patterns in usage of the resources on the site with the aim of providing those resources to users more proactively to ensure they have access to everything they need.
Hughes, Adam G., Stefan D. McCabe, William R. Hobbs, Emma Remy, Sono Shah, and David M. J. Lazer. Forthcoming in Public Opinion Quarterly. “Using Administrative Records and Survey Data to Construct Samples of Tweeters and Tweets.” https://doi.org/10.1093/poq/nfab020.
In review: A paper co-authored with Dr. Kelly McConville advising statistics educators on how to teach basic data wrangling and visualization in R with dplyr and ggplot2 using the Global Volcanism Program’s terrific data.
Alison Rosenzweig and I explored moving a simulated adaptive robot into reality, creating some fun abstract art in the process.
I wrote my senior capstone paper (a kind of expository thesis) about some mathematical aspects of drawing and evaluating districts that make gerrymandering a difficult problem to solve.
Inspired by the 2016 NLM Pill Recognition Challenge, we segmented pills from backgrounds in reference images of pills.
We used C++ and OpenGPL to implement 3-D metaballs.
As an extension of the paper I wrote with Dr. McConville, I presented a virtual poster at the 2017 Electronic Undergraduate Statistics Conference about visualizing map data. This presentation demonstrates and advises undergraduates and educators of undergraduates on how to create this visualizations ranging from simple ggplot2 graphics to R Shiny apps. Code referenced in this presentation is available here.
Peripeteia is an annual event where people from the Swarthmore College community teach short courses on diverse topics. I helped organize this in its inaugural year, as well as the two subsequent years. I also created and maintained the website, and helped organize smaller-scale events related to Peripeteia.
In 2017, I was awarded the celebrated scholarship from AnitaB.org and attended GHC as a Grace Hopper Scholar.
This project won the “Sponsored Prize” at Local Hack Day Swarthmore. It is an SMS service for sending health reminders – our goal was to make the functionality of many fitness tracking smartphone apps accessible to those who do not have or are not comfortable with smartphone technology. This was built using the Twilio API.
An in-progress simple website teaching the basics of bahasa Wooi, a language spoken by around 200 people on Yapen island. This website is an extension of the research I did in 2016 in Manokwari, Papua collecting data on Wooi with the assistance of my language consultant Jimmi Kirihio and my mentor Dr. Emily Gasser.
An R Shiny app visualizing volcano locations and information using the Global Volcanism Program’s data and Leaflet. I have since explored the idea of replicating this using D3.js.
A simple R Shiny app that allows users to explore arrest rates for different demographic groups.