Projects
Dog Adoption Predictor
🐾 Built a LightGBM regression model that scores each dog's likelihood of being adopted quickly on a 0 to 1 scale. Combines tabular pet attributes with VADER NLP sentiment analysis on rescuer descriptions. Trained on 8,132 dogs from the PetFinder dataset achieving a ROC AUC of 0.74.
JPMorganChase Data for Good Hackathon
Collaborated in a 24-hour hackathon, working with a diverse team to expand the reach of nonprofit “Ordinarie Heroes” into a new zip code. Analyzed large-scale datasets to identify key insights and workflows, contributing to the nonprofits expansion efforts. Built a cluster analysis model to enhance the strategic planning and outreach initiatives of “Ordinarie Heroes”
DGI Hackathon 2025
Collaborated with a six-person team to build a data-driven model for Rebuilding Together Nashville that quantifies the community impact of home-repair investments. Analyzed census, housing, and income data to identify “quick-win” repairs that maximize ROI, produced visual storytelling showing 317+ repairs since 2018 and 151 homes restored (2019–2024), and focused on strategies to fight gentrification and preserve affordable housing. Tools: Python, Pandas, Matplotlib, Canva.
YouTube Comment Classifier
Developed a Python program that utilizes machine learning classifiers to categorize YouTube comments from coding tutorial videos into content-related and miscellaneous categories. The project evaluates Naïve Bayes, SVM, Decision Trees, and Random Forest classifiers using accuracy, precision, recall, and F1-score. The dataset, collected via the YouTube Data API, contains 6000 manually labeled comments. 10-fold cross-validation ensures robust evaluation, providing insights into classifier performance for automated comment analysis..
Computer Organization and Architecture - Tutorials
This repository contains tutorials and assignments from my Computer Organization and Architecture course, showcasing my work in C++ and ARM64 Assembly. The C++ tutorials explore topics such as memory management, recursion, and algorithms, while the ARM64 Assembly tutorials focus on low-level programming and understanding the ARM64 architecture.
Movie Recommender System
Developed a Python-based movie recommender system using the MovieLens 100K dataset. Implemented two approaches: one leveraging user similarity to suggest top 10 unwatched movies, and another using clustering to recommend films based on similar users. Extra functionality allows genre-based recommendations. The system analyzes 100,000 ratings from 943 users across 1,682 movies, utilizing machine learning techniques for personalized suggestions.
College Admissions
Developed a Python program that analyzes student data from a CSV file, calculating a weighted score based on GPA, SAT scores, interest levels, and other factors to generate a list of top candidates for admission. It also identifies outliers and checks for trends or discrepancies in academic performance, providing a comprehensive view for fair admissions decisions.
Bayesian Statistics: Fish Market Data Analysis
This project analyzes the Fish Market Dataset to predict fish weight based on continuous variables like Length2 and Height using Bayesian methods. We implement Directed Acyclic Graphs (DAGs), Quadratic Approximation (quap), and Markov Chain Monte Carlo (MCMC) to model relationships between predictors and outcomes. Our approach includes developing Bayesian models, assessing priors, running MCMC simulations, and comparing models using WAIC for predictive accuracy. Visualizations, including counterfactual plots, provide insights into model performance. The dataset, sourced from Kaggle, includes species, weight, and various physical measurements of fish.