Evaluate model performance using intervals (movement data)

Calculate the interval based validation, including predictive distance metric, log likelihood, and log likelihood baseline For distance metrics: What is the distance and probability that BirdFlow model prediction beats the naive S&T probability distribution-based prediction? Get the interval based validation metrics for one transition pair

Usage

get_interval_based_validation_one_transition_pair(
  birdflow_interval_row,
  bf,
  gcd,
  st_dists
)

Arguments

birdflow_interval_row

A row of data in the BirdFlowIntervals object

bf

BirdFlow model

gcd

Matrix of great circle distance

st_dists

Matrix of S&T distribution with weeks as columns, location as rows, probability as values.

Value

A named vector with distance metrics