Sophia Cofone

Within all data, there is a story. I create interactive data visualizations to discover and communicate that story to a variety of audiences. I am skilled at processing and analyzing data using statistical methods to find trends and insights.

I consider myself a life-long learner and a passionate person in and out of work. I enjoy jumping into new project areas, and sharing ideas with my peers. I would like to expand my knowledge of machine learning and statistical analysis, while also improving my current data story-telling abilities.

You will often find me trying new recipes in the kitchen, jamming out to music with friends and family, growing more plants than my apartment can realistically fit, and hiking to the top of a mountain to take in the view.

Semantic image segmentation using Convolutional Neural Networks

I am currently working on a semantic image segmentation project in collaboration with the Biosystems lab at NEU. Using Convolutional Neural Networks (CNNs), we have designed a UNet-inspired CNN model that successfully segments bacteria in biological microscopy images, achieving an impressive IOU score of 92%. See more: WormSPA CNN project paper (coming soon) & GitHub (coming soon).

Gameplay-Based Classification of Valorant Players: Insights and Feature Importance

Can we create data-driven strategies to improve success in Valorant rounds? What are the key gameplay elements that players should prioritize? Is it feasible to categorize players into roles, akin to traditional sports positions? This research project aims to use time-series data from the esports game Valorant to address these questions.

See more: Valorant project Github & presentation slides.

Supervised Machine Learning Applied to Adult vs Baby Face Classification

This project applies classical supervised machine learning models to classify adult vs infant faces using facial coordinate data with over 96% accuracy. Feature engineering, feature selection, and resampling methods are applied to boost recall and accuracy scores as much as possible.

See more: Faces project GitHub.

SoundSynergy: Complete React Web Application

SoundSynergy is a feature-rich web application that allows users to build a music profile, connect with friends that share the same music tastes, and discover new businesses that play music that a specific user will enjoy.

See more: SoundSynergy project GitHub.

Bayes’ Rule Visualization using Spotify data

Have you ever wondered what music you should play at your party to get your friends dancing? This project answers that question using Bayesian model inference and user generated Spotify data.

See more: Bayes’ project GitHub.

Relational Databases using MySQL: Bird-strikes

This project involves building a database to analyze bird strikes on aircraft using an existing data set from the FAA. The main tasks include creating a logical data model and relational schema, implementing the schema in MySQL, loading the data from a CSV file into the database, executing SQL queries, and performing analytics in R.

See more: Relational Databases project GitHub.

Data warehouses using MySQL and SQLite: Publications

This project involves extracting data from an XML document and storing it relationally in a SQLite database. The data is then used to create an analytical database using a star schema in MySQL. The project connects to two different databases simultaneously, which is a common practice for data warehouses.

See more: Data warehouse project GitHub.

Maine Broadband

Broadband internet access greatly varies across the state of Maine. This project visualizes the current state of broadband connectivity using geojson data, and provides a tool for Mainers to see their access level.

See more: Maine Broadband project website & GitHub.

Vaccines

This project investigates the relationship between COIVD-19 vaccines and deaths in the United States, using common data sources from JHU and the CDC.

See more: Vaccines project website & GitHub.

Hi, I’m Sophia. I am a early-career data scientist passionate about leveraging data modeling techniques to narrate the underlying 'story' of data.

Semantic image segmentation using Convolutional Neural Networks

Gameplay-Based Classification of Valorant Players: Insights and Feature Importance

Supervised Machine Learning Applied to Adult vs Baby Face Classification

SoundSynergy: Complete React Web Application

Bayes’ Rule Visualization using Spotify data

Relational Databases using MySQL: Bird-strikes

Data warehouses using MySQL and SQLite: Publications

Maine Broadband

Vaccines

Get in touch with me

Location

LinkedIn

GitHub