projects

A curated selection of data science, machine learning, and research projects.

Projects to Highlight

Business Insights

Machine Learning Focused
Predicting Key Profit Drivers from US E-Commerce Records

Predicting Key Profit Drivers from US E-Commerce Records

Master's capstone at Boston University identifying order-level profit drivers across ~3,300 US e-commerce transactions. A tuned Gradient Boosting model reached a test R² of 0.90 (RMSE $36.81), and a Random Forest classifier separated high- from low-profit orders with 90% accuracy. Discount and sub-category emerged as the key levers for sales and supply chain teams.

View project
Predicting Customer Satisfaction (Olist)

Predicting Customer Satisfaction (Olist)

Master's capstone predicting 1–5 star customer reviews for Olist, a Brazilian marketplace, by merging nine relational tables into ~88,000 orders. Compared KNN regression, Gradient Boosting classification and clustering, and found that seller rating, listing photos and description length matter as much as delivery delays in driving satisfaction.

View project
Automated Property Valuation

Automated Property Valuation

Automated valuation model for residential properties trained on 77,000+ Zillow listings. Engineered size and age features, compared Lasso, Decision Tree and Gradient Boosting, and selected Gradient Boosting with a test MAE of ~$196K, beating the linear baseline by over $40K in cross-validation.

View project

Tools
Colour Palette Extractor Tool

Colour Palette Extractor Tool

A Python tool that gives designers a data-driven starting point for a brand identity. It uses K-Means clustering to pull the dominant colours from any inspiration image, returns their hex codes, and recommends black or white text for each colour based on contrast for accessibility.

View project
Other

Scientific Research

Statistics Focused
Palos Verdes Reef as Refuge Habitat

Palos Verdes Reef as Refuge Habitat

Statistical analysis in R of whether Palos Verdes Reef, an artificial reef in the Southern California Bight, acts as a refuge for sea stars and sea urchins threatened by wasting disease and climate change. Used BACIPS models and generalised linear mixed models to compare species densities before, during and after wasting events, in collaboration with the Vantuna Research Group at Occidental College.

View project