Track Lizard Movement with a Capture-Recapture Model
This example illustrates using a hidden Markov model to reconstruct the most probable lizard migration pattern, as well as estimate a lizard's expected lifespan from a sequence of incomplete trapping records.
Lizards move between three zones represented by states 2, 3, 4. When they die they transition into state 1. Each lizard is initially tagged.
At each time , 25 lizards are captured, each lizard's zone is recorded, and they are released. The recorded 1, 2, or 3 corresponds to an actual zone, while 4 represents a missing value that could mean that the lizard was not captured or that it is dead.
Estimate a 4-state, 4-emission hidden Markov process from the data.
Find the most likely movement pattern for the first lizard.
In the preceding sequence the state 1, death, is an absorbing state.
The mean time to death.