Hidden Markov model for movement data in Ecology

Author

MP Etienne

Published

Dernière Mise à jour Octobre 2024

Course description

Hidden Markov Models are a powerful class of statistical models that are particularly well-suited to analyzing sequential data, where observed patterns are influenced by underlying, unobservable (hidden) states.

In this course, we will explore the basic concepts linked with Hidden Markov Models (HMM):

  • Definition
  • Inference with partially observed data - EM algortihm
  • Baum Welch Algortihm, EM algorithm in the context of HMM
  • Viterbi algorithm to reconstruct Hidden states

The lab session

In ecological research, HMMs have gained prominence for their ability to extract meaningful behavioral states from movement data, such as distinguishing between foraging, resting, and traveling in animals.

We will apply the HMM concepts to analyze movement of animals and we will focus on three Red-footed Boobies (Sula sula). as case studies (trip1, trip2, trip3).

The data used in this lab were collected through the work of Sophie Bertrand (IRD), Guilherme Tavares (UFRGS), Christophe Barbraud, and Karine Delord (CNRS) and i would like to thank the International Associated Young Team (JEAI) IRD Tabasco for providing access to these data.