Andresen's spatial point pattern test

Compares where two point patterns fall over the same areal units (Andresen 2009): for each unit, the difference between its share of the test pattern b and its share of the base pattern a, with a bootstrap p-value from multinomial resamples of each pattern; the S index is the share of units whose difference is not significant (1: the patterns are alike everywhere).

Usage

sptst(a, b, nboot = 200, seed = NULL, alpha = 0.05, scale = 100)

Arguments

a, b

The counts of the base and the test pattern by unit: whole numbers from 0; NA in either leaves the unit out (counted, never taken as 0).

nboot

The number of bootstrap resamples: a whole number from 1.

seed

NULL to draw the simulations' seed from R's generator (so set.seed() fixes them), or one whole number from 0.

alpha

The level of each unit's test, strictly between 0 and 1.

scale

The multiplier of the differences (100: percentage points).

Value

A list: units, a data frame with one row per unit of diff, p and flag (the sign of a significant difference, 0 otherwise; NaN without a p-value); glob, a named vector: units left out, empty in both, eligible and flagged, the S index s, its z against 1 - alpha, the exact p of so many flags and nboot.

References

Andresen, M. A. (2009). Testing for similarity in area-based spatial patterns: a nonparametric Monte Carlo approach. Applied Geography, 29(3), 333-345. doi:10.1016/j.apgeog.2008.12.004

Examples

set.seed(1)
a <- rpois(30, 10)
b <- rpois(30, 10)
sptst(a, b, nboot = 199, seed = 1)$glob
#>        left       empty        elig     flagged           s           z 
#>   0.0000000   0.0000000  30.0000000   2.0000000   0.9333333  -0.4188539 
#>           p       nboot 
#>   0.4464579 199.0000000