John K Best

PhD candidate

Quantitative Ecology & Resource Management

Biography

I am a fourth year PhD student in the Quantitative Ecology and Resource Management program at the University of Washington. My research focuses on applying spatiotemporal statistical models to questions in fisheries management and ecology. In particular, I am developing models for indices of abundance that include fishery-dependent observations. I am also interested in improving the computational efficiency and reliability of Bayesian model fits through the application of advanced Markov chain Monte Carlo tools like the No-U-Turn Sampler. Prior to graduate school I worked in aviation maintenance and the outdoor industry.

Interests

Spatiotemporal statistics
Bayesian statistics
Fisheries management

Education

PhD in Quantitative Ecology & Resource Management, In progress
University of Washington
MS in Statistics, 2016
University of Alaska Fairbanks
BS in Biology & Physics, 2008
Clarkson University

Recent Posts

Down the `rgdal` rabbit hole

Down, down, down I have a package that depends on INLA, which has recently begun failing to load if rgdal isn’t installed correctly. rgdal can’t be installed without the GDAL library, which requires PROJ, which requires SQLite. I needed to install all of these on our cluster without root access. That meant compiling each one to a non-standard location, then ensuring that the next in the sequence knew where this was.

Last updated on Nov 3, 2020 2 min read

Introduction to Hyak

Use Hyak when you need additional resources More processing power Models or data don’t fit in RAM Think carefully about how to use Hyak Do you really need to use it? How can you use it efficiently? Parallelization Which steps are independent? Which steps are bottlenecks? How to minimize overhead? Example: Simulation study Specify operating and estimation models Simulate 100 operating models Fit 300 estimation models Calculate metrics Summarize results Example: Simulation study Specify operating and estimation models Simulate 100 operating models Fit 300 estimation models Calculate metrics Summarize results Which steps are independent?

Last updated on Apr 27, 2020 3 min read

Set up a website for GitHub Pages

Use Git, GitHub, and blogdown to publish an example website on GitHub Pages

Last updated on Jan 28, 2020 7 min read

Compiling R with Intel compilers and MKL on Hyak

Steps to compile custom, optimized R on Hyak

Last updated on Jan 28, 2020 3 min read

Projects

Combining fishery-dependent and -independent CPUE data

Spatially varying catchability for fishery-dependent CPUE standardization.

Preferential sampling in fishery-dependent catch and effort data

Accounting for preferential sampling in spatial models of catch and effort data

Nonstationary spatial processes for fisheries

Incorporating nonstationary spatial processes by allowing for spatially varying covariance parameters.

Best practices for fitting non-linear Bayesian state-space models

Eliminating HMC divergences when fitting Bayesian state-space models for fish populations.

Bayesian models for assessment of Southern Humpback whales

This project is a collaboration between Alex Zerbini, Grant Adams, and myself. We are working to update Zerbini et al. (2006). Additional data are being incorporated to make a more accurate assessment of the population. The software used to fit the model has been almost completely rewritten in order to increase both efficiency and flexibility.