Hailey Ernest
I specialize in cleaning, transforming, organizing, analyzing, and visualizing data, turning complex datasets into clear, actionable insight through statistical modeling, machine learning, and thoughtful storytelling.
Background
About Me
Education
Coursework
Technical Skills
Work
Projects
NASA Capstone: Cryogenic Boiling Detection
Classifies acoustic boiling regimes in cryogenic fuel systems using 70 engineered statistical features, UMAP dimensionality reduction, and HDBSCAN. Achieves 99% clustering stability across 446 acoustic accelerometer runs.
Ondine Biomedical: Data Analytics & Automation
Predictive models, validated data pipelines, and automation tools built as a Data Analytics Intern and Consultant at Ondine Biomedical, spanning high-throughput assay analysis, clinical trial data, and laboratory workflow automation.
WAVES: Wearable Algorithm Research
Research with Cal Poly Kinesiology & Public Health and a WHO-aligned consortium (Cal Poly, NCI, Oxford, Monash). I clean, merge, and validate wrist accelerometer data and test how well open-source algorithms measure MVPA, steps, and sedentary time.
Global Measles Trends Dashboard & R Package
Mixed-effects and logistic regression models in R relating measles incidence to GDP per capita using WHO and World Bank data, packaged as a custom R package and shared through an interactive Quarto and Shiny dashboard.
Coursework
Academic Projects
Exploratory Data Analysis
Two open-ended R analyses: US regional minimum wage trends from 1968–2020, and National Park trail exploration with functional programming and t-test inference.
Databases: Merging & Analysis in SQL
SQL queries from my Database Management Systems course on three datasets (bakery sales, California wine ratings, and the CSU system), using multi-table joins, GROUP BY / HAVING, and nested subqueries.
Shiny App: US Obesity Analysis
Reactive Shiny app exploring obesity rates across all 50 states, filterable by ethnicity, year range, and percentage interval. Features a choropleth US map and time-series for top and bottom states.
Complex Statistical Models: Elderly Falls
Rigorous GLM analysis comparing 8 count regression models on elderly falls data. Every decision documented across 47 log entries, culminating in a Negative Binomial model showing 84.7% fall reduction with exercise.