Tree-Spatial Scan Statistic for Cluster Detection

Implements the tree-spatial scan statistic for detecting clusters that combine both spatial and hierarchical structures, as proposed by Cancado et al. (2025) . The method extends Kulldorff (1997) circular spatial scan statistic and the tree-based scan statistic of Kulldorff et al. (2003) by searching for anomalies in both geographic regions and branches of hierarchical trees simultaneously. The package also provides standalone implementations of Kulldorff's circular spatial scan statistic and the tree-based scan statistic. Statistical significance is assessed via Monte Carlo simulation under a Poisson or binomial model, with optional 'OpenMP' parallelization.

treeSS

Tree-Spatial Scan Statistic for Cluster Detection

Implements the tree-spatial scan statistic (Cançado et al., 2025), which detects clusters that are anomalous in both geographic space and a hierarchical tree simultaneously. The method searches over circular spatial zones and branches of a classification tree to find regions where observed cases significantly exceed expectations under a Poisson or binomial model, selectable via the model argument.

Installation

# CRAN version
install.packages("treeSS")

# Development version from GitHub
# install.packages("remotes")
remotes::install_github("allanvc/treeSS")

Quick start

library(treeSS)

# Example: London road collisions
data(london_collisions)
data(london_tree)

# The scan functions take a data.frame as the first argument and refer to
# its columns by name. This keeps the choice of denominator,
# coordinates, etc. transparent.
result <- treespatial_scan(
  london_collisions,
  cases       = cases,
  population  = population,
  region_id   = region_id,
  x           = x,
  y           = y,
  node_id     = node_id,
  tree        = london_tree,
  nsim        = 999, seed = 42,
  n_cores     = 4L                 # parallelize the MC over 4 threads
)
print(result)

# Extract cluster membership for visualization
cr <- get_cluster_regions(result, n_clusters = 3, overlap = FALSE)

Included datasets

Key functions

• treespatial_scan() — tree-spatial scan (main function)
• circular_scan() — Kulldorff's spatial scan
• tree_scan() — tree-based scan
• filter_clusters() — non-overlapping secondary clusters (Cançado et al. 2025)
• sequential_scan() — sequential adjustment for secondary clusters (Zhang, Assunção & Kulldorff 2010)
• get_cluster_regions() — cluster membership for any visualization package

Visualization

The package is visualization-agnostic. get_cluster_regions() returns a data.frame that can be merged with any spatial object for plotting with ggplot2, leaflet, tmap, or any other mapping package. The bundled sf boundary datasets (rj_map, london_boroughs_map, chicago_map) let the examples map clusters without any external boundary download. See vignette("introduction") for worked examples with ggplot2 + rj_map (Brazil), leaflet + tigris (USA), and leaflet + london_boroughs_map (London).

References

Cançado, A. L. F., Oliveira, G. S., Quadros, A. V. C., & Duczmal, L. (2025). A tree-spatial scan statistic. Environmental and Ecological Statistics, 32, 953–978. doi:10.1007/s10651-025-00670-w