Recursive estimate of central location

Source: R/median_rec.R

Recursive estimate of central location based on depth measures (from the packages depth and ddalpha) or convex body minimizers (package MASS).

median_rec(
  data,
  method = "Projection",
  alpha = 0.5,
  maxIterations = NULL,
  warnings = FALSE
)

Arguments

data

Matrix of numerical values containing the observations (one per row, with two columns for X and Y coordinates)
method

Method to use. Valid options are "MCD" and "MVE" for convex body minimizers, or "L2", "Lui", "Mahalanobis", "Oja", "Projection" (default), "Spatial" and "Tukey" for depth functions
alpha

Proportion of samples trimmed at each iteration (numerical value between 0 and 1, default: 0.5)
maxIterations

Set to a positive integer to limit the number of iterations, to NULL or 0 (default) for no limits
warnings

Logical value, to display the warnings and error raised by the underlying functions

Value

The function returns an S3 object of type BRIL.MedianRec, containing the following values:

median

Coordinate of the recursive median

max

Coordinate of the sample with the highest depth (or the center of the first iteration in the case of convex body minimizers)

iterations

List containing the indices from the samples of data selected at each iteration

References

Adrien Brilhault, Sergio Neuenschwander, and Ricardo Rios - A New Robust Multivariate Mode Estimator for Eye-tracking Calibration - Behavior Research Methods, 2022 - rdcu.be/cI9Pf

See also

plot.BRIL.MedianRec(), print.BRIL.MedianRec(), median_mv(), depth_values(), bril()

Examples


# Illustrative data
XY <- rbind(
  mvtnorm::rmvnorm(300, c(0, 0), diag(2)),
  mvtnorm::rmvnorm(100, c(15, 20), diag(2) * 3 - 1),
  mvtnorm::rmvnorm(150, c(-10, 15), diag(2) * 2 - 0.5),
  mvtnorm::rmvnorm(200, c(5, 5), diag(2) * 200)
)

# Compute the recursive median
res <- median_rec(XY)

print(res)

#> 
#> => Results for median_rec() using "Projection" method (alpha=0.50)
#>   (750 samples, 9 iterations)
#> 
#> Recursive median (median):
#> [1]  0.5549600 -0.1747795
#> 
#> Sample median (max):
#> [1] 0.2865685 1.5814283
#> 
plot(res)