Prepare Data to Plot Cohen's Kappa Correlation Matrix

splnr_get_kappaCorrData() calculates Cohen's Kappa correlation coefficients between a list of prioritizr conservation solutions. The output is a symmetrical matrix suitable for visualizing pairwise agreement using a heatmap.

Usage

splnr_get_kappaCorrData(sol, name_sol)

Arguments

sol

A list of prioritizr solution objects. Each element in the list must be an sf object containing a binary column named solution_1.

name_sol

A character vector providing descriptive names for each solution in the sol list. The length of this vector must match the length of sol. These names will be used as row and column names in the output correlation matrix.

Value

A numeric matrix (matrixOut) representing the Cohen's Kappa correlation matrix between all pairs of solutions. Rows and columns are named according to name_sol.

Details

This function is essential for assessing the similarity or divergence among different conservation plans. It takes a list of prioritizr solution objects, each expected to contain a binary column named solution_1 (indicating selected or unselected planning units).

For every unique pair of solutions in the input list, it computes Cohen's Kappa using the irr::kappa2() function. Cohen's Kappa measures the agreement between two raters (in this case, two conservation solutions) for categorical items, correcting for chance agreement. A Kappa value of 1 indicates perfect agreement, 0 indicates agreement equivalent to chance, and negative values indicate agreement worse than chance.

The resulting matrix is symmetrical, with diagonal elements always equal to 1 (a solution perfectly agrees with itself). This matrix can then be passed to visualization functions like splnr_plot_corrMat() to create a correlation heatmap.

Examples

if (FALSE) { # \dontrun{
# Assuming 'dat_species_bin' is an existing sf object in your package.

# Create a dummy prioritizr problem and solve it for solution 1 (30% target).
dat_problem1 <- prioritizr::problem(
  dat_species_bin %>% dplyr::mutate(Cost = runif(n = dim(.)[[1]])),
  features = c("Spp1", "Spp2", "Spp3", "Spp4", "Spp5"),
  cost_column = "Cost"
) %>%
  prioritizr::add_min_set_objective() %>%
  prioritizr::add_relative_targets(0.3) %>%
  prioritizr::add_binary_decisions() %>%
  prioritizr::add_default_solver(verbose = FALSE)

dat_soln1 <- dat_problem1 %>%
  prioritizr::solve.ConservationProblem()

# Create another dummy prioritizr problem and solve it for solution 2 (50% target).
dat_problem2 <- prioritizr::problem(
  dat_species_bin %>%
    dplyr::mutate(Cost = runif(n = dim(.)[[1]])),
  features = c("Spp1", "Spp2", "Spp3", "Spp4", "Spp5"),
  cost_column = "Cost"
) %>%
  prioritizr::add_min_set_objective() %>%
  prioritizr::add_relative_targets(0.5) %>%
  prioritizr::add_binary_decisions() %>%
  prioritizr::add_default_solver(verbose = FALSE)

dat_soln2 <- dat_problem2 %>%
  prioritizr::solve.ConservationProblem()

# Calculate the Cohen's Kappa correlation matrix between the two solutions.
corrMat <- splnr_get_kappaCorrData(
  sol = list(dat_soln1, dat_soln2),
  name_sol = c("Solution_A_30pct", "Solution_B_50pct")
)
print(corrMat)

# This output can then be directly passed to splnr_plot_corrMat().
# splnr_plot_corrMat(corrMat, AxisLabels = c("Sol A (30%)", "Sol B (50%)"))
} # }