This script simulates binary infection data under an imperfect detection diagnostic test scenario. It provides Stan code to estimate parameters of the regression under imperfect detection (by estimating both sensitivity and specificity of the diagnostic test). Latent discrete infection status is marginalised in the Stan program

Simulate data under a spatially autocorrelated Poisson model and fit a latent Gaussian Process regression in Stan (geostatistical model)

mvgam case study 1: model comparison and data assimilation

Joint species distribution model that uses a Conditional Random Fields generative model for estimating species' associations

In this example we use mvgam to fit a dynamic distributed lag model

In this example we will examine multivariate forecasting models using mvgam. Here we will access monthly search volume data from Google Trends, focusing on relative importances of search terms related to tick paralysis in Queensland, Australia. We will then fit multivariate GAMs in which the latent temporal dependencies are modelled with a dynamic factor process

The development of methods to detect how a species’ occurrence probability covaries with the occurrences of other species (e.g. co-occurrence) is a booming and somewhat divisive area of statistical ecology. Here, I will compare some of the more popular co-occurrence methods against likelihood-based methods that do account for indirect associations using freely-available datasets. The aim of this is to showcase how different co-occurrence analyses work and make recommendations on their use / interpretation