max_method sets how the maximum spatial separation
(spatial_separation_max) is found. The new default,
"exact", searches for the true maximum, which is guaranteed
for Euclidean distances between centroids (see ?uci). The
previous approaches underestimated it by an amount that depends on the
shape of the area; they remain available as "heuristic"
(old default) and "bootstrap", and reproduce earlier
results. The argument bootstrap_border is deprecated.uci() is faster and uses much less memory, especially
with dist_type = "spatial_link" and the bootstrap. Results
are unchanged.parallel now sets the number of threads for
shortest-path distances (dist_type = "spatial_link" only).
The default, FALSE, uses one thread; previously all cores
were used.remove_islands() and argument
uci(remove_islands = TRUE) keep only the largest group of
contiguous polygons, so "spatial_link" works in areas with
islands.uci() now warns when
dist_type = "euclidean" and the study area is markedly
concave or has a large gap inside it (e.g. a bay or a lake), suggesting
"spatial_link" (#11).grid_rio.rds, a hexagonal grid of
Rio de Janeiro with a concave outline and islands.NA, empty or non-polygon geometries, and non-contiguous
polygons with "spatial_link".uci() now accepts columns with NA values.
In such cases, NAs are replaced with zero as they represent
zero activities.spdep::poly2nb(geo, queen=TRUE)) to determine neighboring
polygons whe considering spatial link distances.Major changes
uci function now has a new parameter
dist_type that allows users to choose whether calculations
should be based on "euclidean" distances (Default) or
"spatial_link" distances. Spatial link distances consider
Euclidean distances along the links of spatial neighbour links. In the
case of areas with a concave shape (like a bay), it is strongly
recommended to use "spatial_link" distances (even though
they are computationally more costly) because simple Euclidean distances
can bias UCI estimates in those cases.uci function now has a new parameter
parallel that speeds up calculations for large areas
comprised with many polygons.Minor changes