Follow the links below to jump to the folder and see the code and more information about each project.
- Overview: My team's final capstone project for UC Berkeley Master's in Data Science program won the Hal Varian Award for the best capstone project out of 30+ teams. Using instrumental variables regression, we show the causal impact of PM2.5 (air pollution) on pediatric health outcomes in the state of California. We find that high levels of PM2.5 leads to an increase in cases of pediatric blood vessel disease. See the website showcasing the final results.
- Tools used:
python,PySpark,Tableau
- Overview: Used a decoder transformer model in PyTorch to generate abstractive summaries of long Reddit posts.
- Tools used:
python,PyTorch
- Overview: Designed and carried out an experiment on 500+ individuals to test the impacts of watching a sad video on personality test scores.
- Tools used:
R
- Overview: Combined over 600m records of flight and weather data to predict whether or not a flight would be delayed, 15 minutes before departure time.
- Tools used:
PySpark,SparkSQL,Databricks
- Overview: Created an ensemble regression model to predict the sale price of a home that scored in the top 10th percentile in terms of MSE of all submissions on Kaggle
- Tools used:
python
- Overview: A command line simulator/game of Roulette where you can test strategies, view dataviz of your winning history, and take down the house!
- Tools used:
python
- Overview: A data dashboard built in R Shiny that uses web scrapping and APIs to pull financial data.
- Tools used:
R,Shiny