Projects
Graph Neural Network Based Spotify Recommender
This project was my undergraduate capstone research project that I worked on for 20 weeks with peers Shone Patil and Jiayun Wang.
In this project we wrote scripts to use web apis and query spotify feature information for a large dataset of spotify songs
that appeared in the same playlists we were
using. Then we modeled the data as a playlist coocurence graph with the queried song features and creating a neural network with GraphSAGE architecture to create embeddings
for the spotify songs. Finally we created a multi-layer perceptron network to predict links between different songs and from that we
made a recommender system.
• Webpage
• Report
• Github
COVID-19 Data Challenge in Border Communities
This project was an extracurricular competition that challenged different teams to create a data driven solution to a problem involving COVID-19
in border communitites. My team members and I chose to try to predict likelihood of K-12 schools reopenning in the fall of 2020 using a K-means
clustering method on recent data involving COVID-19 and different features from schools in San Diego County. The process involved research, finding
relevant data, using spatial data joining methods, conducting analytics and feature engineering, and interpreting cluster results into different map
based visualizations with python and ArcGIS.
• Story Board
• Github