Published Papers

Manuscripts in Preparation

  • Quantifying the Association Between Within-Subject Volatility in Serum Albumin and Mortality via a Gaussian Process Model (2017)
    Holsclaw, T., Shahbaba, B., and Gillen, D.L.A
  • Modeling Daily Streamflow with a Periodic Auto-regressive Model with a Gaussian Process Shrinkage Prior (2017)
    Holsclaw, Lall, Smyth.


UCI Team Projects

  • Gaussian process modeling of longitudinal data in a survival context
    Collaborators: Babak Shahbaba and Dan Gillen - UCI Statistics Department
  • Decadal prediction and stochastic simulation of hydroclimate over monsoonal Asia
    Collaborators: Andrew Robertson (PI), Arthur Greene, and Upmanu Lall - IRI at Columbia University
    Supervisor: Padhraic Smyth - UCI, Machine Learning
    Research: Bayesian spatio-temporal stochastic weather generator for rainfall and temperature with multiple inputs
  • Automating Behavioral Coding
    Collaborators: Mark Styvers (UCI) and David Atkins (UW) - Department of Psychiatry and Behavioral Science
    Supervisor: Padhraic Smyth - UCI, Machine Learning
    Research: stochastic modeling of time dependent discrete outcomes for behavioral data
  • Analysis of student behavior in Massive Open Online Courses (MOOC)
    Collaborators: various - Department of Education, UCI
    Supervisor: Padhraic Smyth - UCI, Machine Learning
    Research: analyzing student behavior and predicting outcomes in the MOOC educational setting
  • Breast cancer microarray study using Lasso regression for genetic feature selection
    Collaborators: Bogi Andersen - UCI School of Medicine
    Supervisor: Padhraic Smyth - UCI, Machine Learning

Previous Team Projects
These some projects I worked on as a student:

  • Translating Differential Equation Models Into Kernel Methods for Data Analysis - Phase III (NASA/SJSU) (Large file)
  • Mathematical and Statistical Analysis of Heat Pipe Design (Intel/SJSU)
  • Statistical Analysis of a Non-Linear Chemical Equation Using Numerical Methods (NASA/SJSU)


Dissertation - research done under the advisement of statisticians, Herbie Lee and Bruno Sanso (UCSC). The body of work covers modeling derivative curves with Gaussian processes. This work required extensive research and development of Gaussian process methods applied to inverse problems. A cosmology application drove this statistical research to model dark energy equation of state.

The dissertation was a results of an ISSDM collaboration with Los Alamos National Labs cosmologists Katrin Heitmann, Salman Habib and Ujjaini Alam and statistician David Higdon. We worked with Bayesian parametric and non-parametric models to fit non-linear equations to supernova, baryon acoustic oscilliation, and cosmic microwave backgound radiation data to elicit the form of dark energy. We have exploited the stochastic properties of the Gaussian process to perform analysis for this inverse applied problem.AOAS1009.pdf