gail_mc: Monte Carlo Estimation of Variability from GAIL
In jelsema/GAIL: Geo-Assigment of Irregular Locations

Description Usage Arguments Details Value See Also

For a specified statistic, performs a monte carlo simulation to obtain a distribution based on repeated allocation using gail.

1 2	gail_mc(sp_units, cases, suid, num_cases, max_dist, RAP = gail_rap, unit_value = "m", spat_fun, nsim = 10, seed = NULL, ...)

`sp_units`	Data frame of regular spatial units.
`cases`	Data frame containing the the cases.
`suid`	Column name in both `cases` which contains the spatial unit identification / name.
`num_cases`	Column name from `cases` containing the number of cases, see details.
`max_dist`	The maximum distance at which two locations can be considered neighbors.
`RAP`	Function to be used for calculating assignment probabilities, see details.
`unit_value`	The units of distance, default to meters. See units::set_units.
`spat_fun`	A function to compute the statistic of interest over the spatial units. Can also accept "moran", "localmoran", and "localmoran_full".
`nsim`	The number of Monte Carlo replications to run.
`seed`	If given, sets the seed for the RNG.
`...`	Space for additional arguments (e.g., for gail).

Most of the arguments are passed to the gail arguments of the same names. They are included (instead of using ...) to facilitate understanding this function.

The spat_fun argument performs some computation on the dataset. Currently built-in options are:

spat_fun = "moran" will compute Moran's I using spdep::moran.
spat_fun = "localmoran" will compute the local Moran's I for each spatial unit using spdep::localmoran and return the vector of local Moran's I.
spat_fun = "localmoran_full" will return the full results of spdep::localmoran.

The monte carlo simulation repeatedly re-allocates the cases to the regular spatial units. For each allocation, the result from the argument spat_fun is saved. This could be a scaler (e.g., Moran's I), a vector (e.g. local Moran's I), or something more complex.

A custom function for spat_fun should take the arguments:

dat_values is a vector of numeric data representing the value (number of cases) for each spatial unit
nghb_list is the neighborhood matrix, e.g. as returned by spdep::nb2listw
... for extra arguments. For example, the code of spat_fun for Moran's I is: yyy

   spat_fun <- function( dat_values, nghb_list, ... ){
       spdep::moran( x     = dat_values,
                     listw = nghb_list,
                     n     = length(dat_values),
                     S0    = spdep::Szero(nghb_list), ...)$I
    }