dot-al_egal: Implementation of Exploration Guided Active Learning (EGAL)
In e-sensing/activelearning: Active Learning Satellite Image Time Series

Description Usage Arguments Value Author(s)

This function returns a sits tibble with their score in the metric Exploration Guided Active Learning (EGAL). Samples with a larger EGAL metric should be submitted first to a human expert for classification.

.al_egal(
  s_labelled_tb,
  s_unlabelled_tb,
  sim_method = "correlation",
  alpha = NULL,
  beta = alpha,
  w = 0.5
)

`s_labelled_tb`	A sits tibble with labelled samples.
`s_unlabelled_tb`	A sits tibble with unlabelled samples.
`sim_method`	A character. A method for computing the similarity among samples as described in the package proxy. If the method is registered in proxy::pr_DB as a distance, then the inverse of the distance is used (e.g. 1/distance). proxy::pr_DB as a distance, then the
`alpha`	A double. It controls the radius of the neighborhood used in the estimation of sample density.
`beta`	A double. It controls the radius of the neighborhood used in the estimation of the sample candidate set. A bigger beta gives a bigger set. By default is set to be equal to alpha. If NULL, it is computed internally.
`w`	A numeric (between 0 and 1) only used when beta is NULL. This proportion parameter balances the influence of diversity and density in the selection strategy. When w is 0, EGAL becomes a pure-diversity and when w is 1, EGAL becomes a pure density-based sampling algorithm.

           A sits tibble with the EGAL metric. This metric
                   ranks samples based on their density and diversity.
                   Those samples with highest EGAL should be selected
                   first for labeling.