Data Science Student
12-week intensive project-focused online data science bootcamp. Self-designed data science projects from conception to presentation. Topics covered include data collection, data management, exploratory data analysis, modeling with machine learning, and data visualization. Projects completed include:Predicting In Game SSB Melee Win Probabilities:* Calculated statistics on more than 18,000 games of Super Smash Bros. Melee using community-written tools and self-developed algorithms* Created Tensorflow and Scikit-learn classifiers to predict in game win probabilities at any point in amatch, achieved average log loss of 0.25 and ROC AUC of 0.98* Developed and deployed interactive Streamlit app where users can upload a game and see odds overtime and other game statisticsWheel of Time NLP:* Used NLTK, TfidfVectorizer, and NMF to topic model main 14 Wheel of Time books* Constructed undirected graphs of character interaction and created interactive network visualizations* Built interactive Streamlit app to visualize findings of topic modeling and network analysisExoplanet Classification:* Utilized tuned XGBoost model to classify objects of interest catalogued by the Kepler SpaceTelescope as candidate exoplanets with 91% recall* Created interactive Tableau dashboard of Kepler Objects of Interest, including findings of modeling processBoard Game Popularity Prediction:* Used BeautifulSoup to scrape data on 10,000 board games from BoardGameGeek.com* Programmed linear regression with LASSO regularizations models to predict the popularity of boardgames with final R^2 of 0.49
