Ken Nakatsu's Portfolio

Applied Mathematics and Statistics Emory University Undergraduate with interest in data science, mathematical statistics, and actuarial science. Extensive applications of data science concepts to cancer biology research with the John A. Burns School of Medicine at the University of Hawaii Manoa.

sRNAfrag - Analyzing Small RNA Fragmentation with a Graph Theory Approach

In this project, I utilized graph theory in a pythonic implementation to cluster fragments from sequencing data. Coupled with over 50 functions to process and analyze sequencing data realted to this method, the project serves as an avenue to increase our understanding of small RNA fragmentation which has been shown to be relevant to disease.

GeoHazard-GraphDB: A Neo4j Product Mapping the Relationships of Hazardous Chemicals with Their Health Effects

In "GeoHazard-GraphDB," I spearheaded the creation of a Neo4j graph database to elucidate the complex interconnections between hazardous chemicals and their health effects across Georgia, integrating diverse data sources for impactful environmental insights and user-friendly data visualization.

Improving Random Forest Lung Cancer Diagnostic Model Performance in Minority Populations

In this bioinformatics project, I analyzed 215 lung cancer small RNA sequencing samples using DESeq2 and implemented a Random Forest model to classify health states, achieving strong performance metrics across diverse populations. Given the importance of equitable diagnostic models, I discovered that implementing another set of features improved the diagnostic model for minority populations.

Predictive Modeling of Disease on the Population and Individual Level

For the Emory Data Science Club, I created a model to predict breast cancer incidence based on small RNA sequencing data utlizing random forest models. With the University of Hawaii Data Science Institute, I created a project predicting disease incidence with support vector machines, multi-varaible regression, and regression trees based on weather variable.

Locally Hosted Large Language Model Tutors for Self-Learners

This project utilizes a LLAMA 13b parameter model and a vectorized database to provide more informed responses to users seeking to learn on their own with textbooks. I would love to extend its utility to underserved high schools.

Mathematical Applications

With every math course I take, I seek to make one application or visual for at least one topic in the course. Here, I demonstrate my ability to apply seemingly arbitrary topics or explain the concept through fun visuals. Currently a work in progress.

Multi-Variable Calculus

Depicts the drawing of parametric graphs, parametric derivatives, partial derivatives, contour or level curves, and the tangent plane.