a few course and independent projects. reports open in google drive where linked.
python · scikit-learn · pandas
Compared random forest and neural network models on structured NYC trip data to predict duration and tipping, handling heavy skew and extremes. Trip distance dominated duration prediction; tip classification stayed strong.
python
Game-theoretic view of conversation as a repeated game, with 450 turn-level observations across 30 dialogues. Personal disclosure rose with reciprocity and adaptive trust over time.
python
Reddit sentiment features to forecast next-day SHIB returns with an LSTM, then a Markowitz-style layer to choose SHIB–USDC weights from risk aversion, forecasts, and volatility.
more coming as i finish things.