Bayesian Markov model with Polya-Gamma sampling for estimating individual behavior transition probabilities from accelerometer classifications

Abstract

The use of accelerometers in wildlife tracking provides a fine-scale data source for understanding animal behavior and decision making. We modeled transitions between flying, feeding, stationary and walking behavior states using a first-order Bayesian Markov model, introducing Polya-Gamma latent variables for automatic sampling of covariate coefficients from the posterior distribution. Our model provides a unifying framework for including both acceleration and GPS data, with straightforward inference of behavioral time allocation across used habitats.

Publication
Journal of Agricultural, Biological and Environmental Statistics, (25), pp. 365–382, https://doi.org/10.1007/s13253-020-00399-y