Define the unit of analysis
Distinguish a participant, trip, flight leg, and measurement. Build keys and eligibility rules before counting records or fitting models.
BIOSTATISTICS · DATA SCIENCE · RESEARCH SOFTWARE
Turning complex research data into defensible decisions.
I build statistical workflows that make the reasoning visible: how records are joined, how outcomes are defined, how models are checked, and where a result stops being certain.
Research portfolio based on reviewed university project materials. Raw participant data is excluded.
01 / DEFINEOutcome & cohortEligibility, baseline, censoring
03 / VERIFYDiagnostics & checksAssumptions, residuals, reconciliation
METHODS FIRST ↗
01Data integration
02Statistical modeling
03Interactive analysis
04Validation & reporting
SELECTED WORK / 01—06
Research, explained from the inside.
Each case study shows the question, the technical choices, the decision logic, and the checks that make the result interpretable.
THE ANALYTICAL METHOD
Logic before output.
A credible analysis is a chain of decisions. I make each link inspectable, from the first source file to the final interpretation.
Make time and context explicit
Normalize timestamps and time zones; join records across trips; derive sleep, performance, and schedule windows using documented boundaries.
Choose a method for the question
Use survival analysis for time to recovery, mixed models for repeated observations, and descriptive comparisons when causal claims are unsupported.
Check the answer
Review missingness, residuals, proportional hazards, influential records, and source-to-output reconciliation before presenting a conclusion.
TOOLS & CAPABILITIES
The stack behind the work.
Project technologies are separated from methods, so a tool name never stands in for an analytical result.
Analysis & modeling
R · survival · lme4 · lmerTest · car · DHARMa · effectsize
Data preparation
dplyr · tidyr · purrr · lubridate · stringr · readxl · openxlsx
Visualization & applications
ggplot2 · Plotly · Shiny · shinydashboard · DT · R Markdown
Methods
Cox models · censoring · mixed-effects models · feature engineering · cohort checks · validation
SELECTED CERTIFICATIONS
Survival Analysis in R · SQL for Data Science · Intermediate Python · Joining Data with pandas
DataCamp, University of California Davis. Course credentials are shown separately from technologies evidenced in the research projects.LET’S TALK RESEARCH
Have a question about the methods?
These case studies make the project logic and evidence available for discussion without exposing participant-level research data.