Research Paper: Online Drug Markets
A 15-page group research paper on how darknet drug markets use cryptocurrency and escrow systems, and how blockchain analytics help law enforcement trace them. Group project, 2023.
Read the PaperProjects
Data analysis, signal processing, and dashboard projects in Python, R, SQL, and Tableau. Each one covers the question, the methods, and what I found.
Python · Signal Processing
Heart Rate and Step Counting from Phone Sensors
Estimated heart rate and breathing rate from a fingertip held over a phone camera, and counted steps from raw accelerometer data. Group project, 2023.
- Built a high-pass Butterworth filter to separate the heartbeat from breathing and baseline drift
- Detected peaks to count beats and convert them to beats per minute
- Used notch and low-pass filters to recover the slower breathing signal
- Turned three-axis accelerometer readings into one orientation-independent signal for step counting
Tools: Python, NumPy, SciPy, Matplotlib, Jupyter
R · GIS
Mapping Cropland by Watershed
Measured how much land was planted with corn and soybeans inside 52 watersheds in the Great Lakes and Upper Mississippi basins, using USDA satellite crop maps. In 2009 the watersheds held about 20,600 km² of corn and 9,200 km² of soybeans.
- Loaded watershed boundary shapefiles and reprojected them to match the crop rasters
- Summed the exact pixel area of each crop inside every watershed boundary
- Set up the calculation to repeat for every year from 2009 to 2020 and export a tidy dataset
Tools: R, sf, raster, exactextractr, USDA Cropland Data Layer
Python · Exploratory Analysis
Social Capital in Massachusetts
Explored the Social Capital Atlas, a county-level dataset on friendships across income groups, volunteering, and civic life. I focused on whether Massachusetts counties with more volunteering also have more civic organizations.
- Clustered all U.S. counties on civic engagement measures with k-means
- Narrowed to the 14 Massachusetts counties for closer comparison
- Found a weak positive relationship (R² = 0.19), with Nantucket and Dukes as clear outliers
Tools: Python, pandas, scikit-learn, statsmodels, seaborn
Tableau · Business Intelligence
Sales and Profit Dashboards
Built a set of Tableau views on a retail sales dataset to answer business questions: which categories earn the most, which products sell best, and which cities lose money.
- Created a calculated profit margin field to rank cities by profit-to-sales ratio
- Charted profit by category and found the top product sub-categories and best-selling products
- Tracked yearly profit for each country from 2011 to 2014
Tools: Tableau Desktop, calculated fields, dashboards
R · Regression
What Predicts a Developer's Salary?
Used the 2021 Stack Overflow Developer Survey (83,439 responses) to test which background factors predict yearly pay. Team project, 2022.
- Cleaned self-reported salary data and removed extreme outliers
- Compared linear models with stepwise regression to select predictors
- Stepwise regression kept age as the one significant predictor, but both models had high prediction error, a useful lesson about noisy survey data
Tools: R, ggplot2, linear and stepwise regression
Read the ReportMore Work
Research writing and work in progress.
(VR)2 Dashboards
Interactive dashboards that track the (VR)2 virtual reality pre-employment training program at Viability. A public version is in progress.
Want to Talk About a Project?
I'm happy to walk through any of this work, or discuss how I could help with yours.