Robust bandwidth estimation for kernel density estimation — kde_bandwidth • weird

Robust bandwidth estimation for kernel density estimation

Usage

kde_bandwidth(
  data,
  method = c("robust_normal", "double", "lookout"),
  max.iter = 2
)

Arguments

data: A numeric matrix or data frame.
method: Method to use for selecting the bandwidth. robust_normal uses a robust version of the normal reference rule. lookout uses the topological data analysis approach that is part of the lookout algorithm.
max.iter: How many times should the lookout method be iterated. That is, outliers (probability < 0.05) are removed and the bandwidth is re-computed from the remaining observations.

Value

A matrix of bandwidths (or scalar in the case of univariate data).

Author

Rob J Hyndman

Examples

# Univariate bandwidth calculation
kde_bandwidth(oldfaithful$duration)
#> [1] 4.526927
# Bivariate bandwidth calculation
kde_bandwidth(oldfaithful[,2:3])
#>           [,1]      [,2]
#> [1,]  32.85912   182.179
#> [2,] 182.17895 16357.846