Remove NAs from Spatial Data Using Nearest Neighbour

splnr_replace_NAs() replaces missing (NA) values in a specified column of an sf dataframe with the value from the nearest spatial neighbor.

Usage

splnr_replace_NAs(df, vari)

Arguments

df

An sf dataframe. This dataframe must contain a geometry column and the vari column with potential NA values.

vari

A character string specifying the name of the column in df from which NA values are to be removed and replaced. This column must exist in df.

Value

An sf object identical to the input df, but with NA values in the vari column replaced by values from their nearest non-NA neighbors. If no NAs are found, the original df is returned unchanged.

Details

This function is useful for imputing missing data in spatial contexts. It identifies all Planning Units with NA values in the vari column. For each of these, it finds the geographically closest Planning Unit that does not have an NA value in vari, and then copies that non-missing value. This approach leverages the spatial autocorrelation often present in environmental and species data.

The st_nearest_feature() function from the sf package is used for determining the closest neighbor.

Examples

if (FALSE) { # \dontrun{
# Assuming 'dat_species_prob' is an existing sf object in your package.
# For demonstration, let's artificially introduce some NAs in 'Spp2'.
df_with_na <- dat_species_prob %>%
  dplyr::mutate(Spp2 = ifelse(runif(n()) < 0.2, NA, Spp2))

# Replace NAs in 'Spp2' using nearest neighbor imputation.
df_no_na <- splnr_replace_NAs(df = df_with_na, vari = "Spp2")
print(sum(is.na(df_no_na$Spp2))) # Should be 0 if successful
} # }