Skip to contents

Repair spatially overlapping adjacent putative clusters

Source: R/2_repair.R
spatial_repair.Rd

Iteratively merges temporally adjacent putative clusters that overlap spatially. This operates on the .putative_cluster column created by slice_ctdf() and updates it in-place.

Usage

spatial_repair(ctdf, time_contiguity = TRUE)

Arguments

ctdf

A ctdf object. Must contain .putative_cluster (typically produced by slice_ctdf()).

time_contiguity

Logical; if TRUE, enforce temporal contiguity within clusters by filling internal gaps as described above. Default is TRUE.

Value

The input ctdf, invisibly, with .putative_cluster updated in-place.

Details

For each adjacent pair, the locations in each cluster are combined separately and their convex hulls are tested for intersection. If the convex hulls do not intersect, a k-nearest-neighbour (kNN) mixing test is used as a secondary criterion. The kNN test pools the two clusters' locations, uses floor(sqrt(n)) neighbours and compares the observed fraction of neighbours assigned to the other cluster with the expected mixing fraction under random labels. The pair is considered overlapping when the observed mixing is at least 60% of that expectation.

If time_contiguity = TRUE, missing .putative_cluster values between identical forward- and backward-filled labels are filled, so each cluster becomes temporally contiguous (short spatial outliers inside a cluster are absorbed).