Photo of Chris

Chris Comiskey, PhD

Cell: 303-638-2401
Email:

I am a statistician, data scientist, and R programmer with problem-solving, critical thinking, and communication skills. I carefully assess and articulate project goals, process data with tidyverse packages, conduct appropriate statistical analysis, create visuals with ggplot2, write reports with R Markdown and LaTeX, and present results with Beamer slides.

Skills

Statistical Analysis

  • Bayesian methodologies, including hierarchical models
  • Generalized linear models, including logistic regression
  • Design and analysis of experiments
  • Spatial statistics, including Gaussian Random Fields
  • Survival analysis, including Cox Proportional-Hazard models
  • Time series, including autocorrelated hydrologic series
  • Machine learning, including classification with decision trees
  • Graphics, plots, visualizations with ggplot2
Experience

Data Scientist Open Data Group, 9/2017 - Present

  • Predict mechanical failure probabilities with survival analysis decision trees
  • Predict macroeconomic indicators with ARIMA time series modeling
  • Clean, process, and wrangle messy big data, with tidyverse packages
  • Scrape web data with rvest, httr, jsonlite, etc.
  • Develop R packages for clients, using devtools, roxygen2
  • Write R Markdown reports, give LaTeX Beamer presentations
  • Develop R SDK components for use with ODG software REST API
  • Conduct literature reviews of latest machine learning techniques
  • Use Git, GitHub, Bash, MySQL, Python, HTML, Docker

Research OSU, 2013 - 2017

  • Developed variable-resolution heat maps for spatial data
  • Developed R package varyres for implementing variable-resolution heat maps
  • Developed interactive heat map confidence intervals for spatial estimators
  • Developed R package mapapp to create interactive heat map confidence intervals with RStudio’s Shiny
  • Modeled spatially correlated Bernoulli random variables with Bayesian hierarchical logistic regression (with GRF) model
  • Estimated “Effective Sample Size” for highly autocorrelated hydrologic time series, using AR(1) models
  • Thesis: "Take me out to (analyze) the ballgame : Visualization and analysis techniques for big spatial data"

Teaching 2010 - 2011, 2013 - 2017

R

Additional Tools of My Trade

  • dplyr for data manipulation
  • shiny for interactive-visual web applications
  • stringr, readr for string manipulation
  • lubridate for working with dates
  • rstan for Bayesian modeling
  • fields for spatial statistics
  • spBayes for spatial Bayesian modeling
  • INLA for numerical approximations in Bayesian modeling
  • mapapp for interactive heat map confidence intervals (aut)
  • varyres for creating variable resolution heat maps (aut)
  • rpart, LTRCtrees for survival analysis trees
Education

PhD, Statistics 2017

Oregon State University

M.S., Statistics 2014

Oregon State University
Provost's Distinguished Graduate Fellow

B.S., Mathematics 2007

University of Colorado Denver

Supplementaries

References

Websites