Local Getis-Ord Gi* statistic per neighbourhood

Returns Gi* per neighbourhood (count vector aggregated from the incident data.frame), using a binary k-NN spatial weights matrix with self-inclusion (Gi* convention). z-score interpretation: Gi*1.96 = significant hot spot at alpha=0.05; Gi* < -1.96 =significant cold spot.

Usage

morie_tps_getis_ord_g_star(
  df,
  ds_name = "?",
  hood_col = "HOOD_158",
  k_neighbours = 5L,
  top_n = 20L,
  lat_col = "LAT_WGS84",
  lon_col = "LONG_WGS84"
)

Arguments

df

Incident-level data.frame.

ds_name

Tag for the result title.

hood_col

Neighbourhood id column.

k_neighbours

k for the (binary) k-NN weights graph.

top_n

Number of top hot/cold spots to surface.

lat_col, lon_col

WGS84 column names.

Value

A named list with Gi* per hood, the top hot/cold spot tables, and tallies of hot/cold spots at alpha=0.05.

Examples

set.seed(2026)
df <- data.frame(
  HOOD_158 = sample(letters[1:20], 400, replace = TRUE),
  LAT_WGS84 = 43.6 + runif(400, 0, 0.2),
  LONG_WGS84 = -79.4 + runif(400, 0, 0.2)
)
morie_tps_getis_ord_g_star(df)
#> Getis-Ord Gi* -- ?
#> ==================
#> Call: morie_tps_getis_ord_g_star(df=<400r>, hood_col=HOOD_158, k=5) 
#> 
#>   Spatial unit              HOOD_158
#>   Neighbourhoods            20
#>   k-NN (incl. self)         6
#>   Hotspots (Gi* > 1.96)     2
#>   Cold spots (Gi* < -1.96)  0
#>   Max Gi*                   2.3321
#>   Min Gi*                   -1.1154
#> 
#> Across 20 neighbourhood(s) with a binary k=6-NN weights matrix (self-inclusive), Gi* identifies 2 hot spot(s) (Gi* > 1.96) and 0 cold spot(s) (Gi* < -1.96) at alpha=0.05. Max Gi* = +2.33 at 'm'; min Gi* = -1.12 at 'q'.