Test residuals for spatial autocorrelation — plot_residual

Computes Moran's I test for spatial autocorrelation in calibration residuals. Significant spatial autocorrelation suggests model misspecification — the IDW kernel may be missing spatial structure.

Usage

plot_residual_autocorrelation(
  result,
  summary_fn = "rmse",
  residual_type = c("raw", "pearson", "deviance"),
  nsim = 999L,
  ...
)

Arguments

result: An interpElections_result object.
summary_fn: Summary function for collapsing residuals across brackets: "rmse" (default), "mean", "mean_abs", "max_abs".
residual_type: Residual type: "raw", "pearson", or "deviance". Default: "raw".
nsim: Integer. Number of permutations for the test. Default: 999.
...: Ignored.

Value

A ggplot object (invisibly). Prints a Moran scatterplot and the test result. Returns NULL with a message if spdep is not installed.