I answer
questions
with data.
Most of the job is working out what the data can’t tell you.
“Does the landscape change how fast it spreads?”
\[\mu_i = f(\rho_{0,i}) + f(\rho_{1,i}) + f(\rho_{2,i})\]
\[\rho_{j,i} = T + \left(\frac{1}{\zeta_{j,i}}\right) \times D_{j,i} \quad \text{for } j \in \{0, 1, 2\}\]
\[D_{j,i} = -\sqrt{(x_j - x_i)^2 + (y_j - y_i)^2}\]
\[\zeta_{j,i} = \beta_0 + \class{eq-hi}{\beta_1} \times z_i\]
The question decides the model, not the other way round.
I generally work with messy data, collected, scraped, or simulated, and often not actually measuring what we hope it does.
Interdisciplinary, but a lot in biology and medicine.
No favourite method. Mostly hierarchical, Bayesian, and simulation-based models, chosen to suit the question.
Statistics courses at undergraduate and postgraduate level, focused on building genuine understanding rather than recipe-following.
BI3010, Statistical Analysis of Biological Data
From linear models through to GLMs. The goal is understanding the theory well enough to apply it to your own questions, with R and ggplot2 throughout.
PGR, Generalised Linear Models
A free course on the theory and application of GLMs, open to students from any university and any discipline, part of a wider series covering R and linear models.
Detailed walkthroughs of statistical models, methods, protocols, and concepts, mostly written for students working with me during their honours projects, but available to anyone who wants them.
MODELSIntroduction to GLMs
Theory and application of GLMs, from model structure to interpretation. Part of the PGR series.
MODELSBayesian Occupancy Models
Occupancy modelling for presence/absence data with imperfect detection, theory and R.
MODELSStructural Topic Models
STMs for text analysis, written to be accessible without a machine-learning background.
MODELSRobust Design Mark-Recapture
Estimating population size and demographic parameters from capture-history data.
PROTOCOLCamera Trap Pipeline
End-to-end protocol for the NE Scotland project: field recording to camtrapR detection histories.
TEACHING AIDStatistical Distributions
Interactive explorer with adjustable parameters, to build intuition for how distributions behave.
PhD students I currently supervise, and the undergraduate honours theses I’ve supervised.
Integrated population models of tawny owls.
Geoprofiling methodology development.
Predicting vole population cycles using raptor data.
Refining spatial methodology for marine environments.
Currently Lecturer in Applied Statistics, University of Aberdeen, previously Senior Statistical Ecologist at the Animal & Plant Health Agency, and Teaching / Research Fellow at Aberdeen. PhD Ecology (rodent pest population dynamics) and a First-Class BSc Zoology, both Aberdeen.