Starting to work on the Vesuvius Challenge

I’ve been developing an autosegmentation tool as part of the Vesuvius Challenge. It uses DBSCAN for initial segmenting, and I’m currently working on implementing a step to correct low-confidence regions according to data from high-confidence regions using Bayesian inference.

An n=1 trial on the effects of Vyvanse

I ran a 30-day trial to figure out the effect my ADHD medication has on my productivity and mood.

Lead-acid battery recycling and Dodd-Frank 1502

This is a piece I wrote about the relationship between demand-side regulation like Section 1502 of the Dodd-Frank act and lead policy.

Predicting Population Size

This is a program I wrote in Python to answer questions of the type “If I sample from a pool 100 times, and 1 object appeared 3 times, 2 objects appeared twice, and every other object appear once, what is the most likely size of the pool?” (which I later learned biologists know as the mark and recapture process to estimate population sizes). It uses a Monte Carlo algorithm and recursive search to estimate the most likely pool size your sample came from.

The Angel Problem

I got an undergraduate research grant to study the angel problem under Nadia Lafrenière and Assaf Shani. I’ve developed an interactive tool in Python to explore the problem and illustrate parts of the proof.