Machine Learning Projects
Cryptocurrency Price Predictions
This was my final project for the Machine Learning Engineer Bootcamp at UC San Diego. I used several previous years of time series data of cryptocurrency prices, along with other data points commonly associated with stocks, to create several models to predict future prices. The final product was an automated API that updated predictions to a website, which included the vanilla Transformer model for the 2017 paper “Attention is All You Need”, a fairly generic LSTM model, and a model from a 2022 paper titled “Are Transformers Effective for Time Series Forecasting?” which the author called “DLinear”. What I found is that the DLinear did perform slightly worse than the Transformer model, however it could be trained incredibly quickly, and also took up much less space. Please see the below Github link for more information on the project.
https://github.com/camdenhine/Cryptocurrency-predictions
The website is no longer updated regularly, but please see the below site for a snapshot of the website last year.
https://cryptocurrencies-price-predict.herokuapp.com/
Bootcamp Projects
These were several graded assessments completed during the Machine Learning Engineering Bootcamp, which I completed in December 2022. Among these projects were problems such as data wrangling, webscraping, linear and logistic regression, and also problems such as anomaly detection, recommendation systems, and time series forecasting. In these problems I used many different Python libraries to train, test, and evaluate many common ML models used in the modern day. Please see the below Github link for further information on each of the individual projects.
Machine Learning on Google Cloud (Coursera)
These were several courses, including some hands-on projects, that are meant to prepare you for the Google Cloud Machine Learning Engineering Certification. Among these include using BigData, storing and loading models, Tensorflow and TFX, Feature Engineering, along with many other topics. Please see the certificate, as well as the Github repository that contains some of the projects.