nb_match_test: Local Neighbor Match Test

In JosiahParry/sfdep: Spatial Dependence for Simple Features

Local Neighbor Match Test

Description

Implements the Local Neighbor Match Test as described in Tobler's Law in a Multivariate World (Anselin and Li, 2020).

Usage

nb_match_test(
  x,
  nb,
  wt = st_weights(nb),
  k = 10,
  nsim = 499,
  scale = TRUE,
  .method = "euclidian",
  .p = 2
)

Arguments

x	a numeric vector or a list of numeric vectors of equal length.
nb	a neighbor list object for example as created by st_contiguity().
wt	a weights list as created by st_weights().
k	the number of neighbors to identify in attribute space. Should be the same as number of neighbors provided in st_knn.
nsim	the number of simulations to run for calculating the simulated p-value.
scale	default TRUE. Whether x should be scaled or not. Note that measures should be standardized.
.method	default "euclidian". The distance measure passed to stats::dist().
.p	default 2. The power of Minkowski distance passed to the p argument in stats::dist().

Value

a data.frame with columns

• n_shared (integer): the number of shared neighbors between geographic and attribute space

• nb_matches (list): matched neighbor indexes. Each element is an integer vector of same length as the ith observation of n_shared

• knn_nb (list): the neighbors in attribute space

• probability (numeric): the geometric probability of observing the number of matches

• p_sim (numeric): a folded simulated p-value

Examples

if (requireNamespace("dplyr", quietly = TRUE)) {
library(magrittr)
guerry %>%
  dplyr::transmute(nb = st_knn(geometry, k = 10),
         nmt = nb_match_test(list(crime_pers, literacy, suicides),
                             nb, nsim = 999)) %>%
  tidyr::unnest(nmt)
 }

JosiahParry/sfdep documentation built on Sept. 7, 2024, 6:15 a.m.