Depth functions wrapper

Source: R/depth_values.R

Computes the depth values with respect to the distribution provided in data of either all the coordinates given in u, or all observations from data if u is not provided. Depth computations rely on the packages depth and ddalpha.

depth_values(data, u = NULL, method = "Projection", warnings = FALSE)

Arguments

data

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

Matrix of numerical values containing the coordinates for which depth values are to be computed (in rows, with two columns for X and Y coordinates). When missing or NULL, depth values will be computed for all observations from data
method

Depth function used. Valid options are "L2", "Lui", "Mahalanobis", "Oja", "Projection" (default), "Spatial", "Tukey"
warnings

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

Value

Array of numerical values containing the depth of each observation from data, or from u if provided (these values are all set to 0 in the occurrence of errors)

See also

bril(), filter_outliers(), median_rec(), median_mv()

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 depths
D <- depth_values(XY, method = "L2", warnings = TRUE)

# Plot distribution with depth color scale
plotColors <- colorRampPalette(c("maroon4", "steelblue4", "green4", "gold"))(20)
plot(XY, pch = 20, asp = 1, col = plotColors[as.numeric(cut(D, breaks = 20))],
     xlab = "X", ylab = "Y")


# Plot depth values
plot(1:nrow(XY), D, pch = 20, col = plotColors[as.numeric(cut(D, breaks = 20))],
     xlab = "Index", ylab = "Depth")


# Compute depth for a single point
depth_values(XY, c(10, 3), method = "L2")

#> [1] 0.3940824

# Compute depth for three sets of coordinates
depth_values(XY, rbind(c(10, 3), c(65, 8), c(0, 1)), method = "L2")

#> [1] 0.3940824 0.1395141 0.4590793