optimal spatial data discretization based on SPADE q-statistics

Function for determining the optimal spatial data discretization based on SPADE q-statistics.

Usage

cpsd_disc(
  formula,
  data,
  wt,
  discnum = 3:8,
  discmethod = "quantile",
  strategy = 2L,
  increase_rate = 0.05,
  cores = 1,
  return_disc = TRUE,
  seed = 123456789,
  ...
)

Arguments

formula: A formula of optimal spatial data discretization.
data: A data.frame or tibble of observation data.
wt: The spatial weight matrix.
discnum: (optional) A vector of number of classes for discretization. Default is 3:8.
discmethod: (optional) The discretization methods. Default all use quantile. Noted that robust will use robust_disc(); rpart will use rpart_disc(); Others use sdsfun::discretize_vector().
strategy: (optional) Discretization strategy. When strategy is 1L, choose the highest SPADE model q-statistics to determinate optimal spatial data discretization parameters. When strategy is 2L, The optimal discrete parameters of spatial data are selected by combining LOESS model.
increase_rate: (optional) The critical increase rate of the number of discretization. Default is 5%.
cores: (optional) A positive integer(default is 1). If cores > 1, a 'parallel' package cluster with that many cores is created and used. You can also supply a cluster object.
return_disc: (optional) Whether or not return discretized result used the optimal parameter. Default is TRUE.
seed: (optional) Random seed number, default is 123456789.Setting random seed is useful when the sample size is greater than 3000(the default value for largeN) and the data is discretized by sampling 10%(the default value for samp_prop in st_unidisc()).
...: (optional) Other arguments passed to st_unidisc(),robust_disc() or rpart_disc().

Value

A list with the optimal parameter in the provided parameter combination with k, method and disc(when return_disc is TRUE).

x: discretization variable name
k: optimal number of spatial data discreteization
method: optimal spatial data discretization method
disc: the result of optimal spatial data discretization

Note

When the discmethod is configured to robust, it will operate at a significantly reduced speed. Consequently, the use of robust discretization is not advised.

References

Yongze Song & Peng Wu (2021) An interactive detector for spatial associations, International Journal of Geographical Information Science, 35:8, 1676-1701, DOI:10.1080/13658816.2021.1882680

Author

Wenbo Lv lyu.geosocial@gmail.com

Examples

data('sim')
wt = sdsfun::inverse_distance_swm(sf::st_as_sf(sim,coords = c('lo','la')))
cpsd_disc(y ~ xa + xb + xc,
          data = sim,
          wt = wt)
#> $x
#> [1] "xa" "xb" "xc"
#> 
#> $k
#> [1] 5 5 7
#> 
#> $method
#> [1] "quantile" "quantile" "quantile"
#> 
#> $disv
#> # A tibble: 80 × 3
#>       xa    xb    xc
#>    <int> <int> <int>
#>  1     1     4     3
#>  2     4     4     6
#>  3     2     5     3
#>  4     1     3     2
#>  5     4     4     5
#>  6     2     4     4
#>  7     4     4     6
#>  8     1     2     2
#>  9     4     3     4
#> 10     5     1     6
#> # ℹ 70 more rows
#>