I’m working with Professor Steinert-Threlkeld from UCLA’s Public Affairs department to analyze if a superlinearity exists between city population and protest sizes. We are using ACLED dataset for protests across the world and Geonames for population. So far, no correlation has been found between the 2 variables. Feel free to check out the GitHub repo.
In this project, I explored Oil Prices for the past 6 years (source) utilizing ARIMA and an state-of-the-art time series linear model. Check out my presentation here for to see the results. My ARIMA model achieved an MSE of 5.88 and < 0.01 for LSTF. Here’s the code for ARIMA models: GitHub repo.
In this group project I led, we tried to predict K-Pop song’s popularity through machine learning and natural language processing. We were first inspired by the BTS 147 Songs Audio Features (Spotify) dataset from Kaggle, but to conduct our analysis, we referred to the following two datasets: the Kpop Artists and Full Spotify Discography and BTS Lyrics dataset (more details in later sections). Before proceeding, we extracted the “Popularity” scores Spotify generated for each song we observed through the Spotify API and appended it to the datasets mentioned above....