weighted_quantile: Weighted Quantiles

In MetricsWeighted: Weighted Metrics and Performance Measures for Machine Learning

Weighted Quantiles

Description

Calculates weighted quantiles based on the generalized inverse of the weighted ECDF. If no weights are passed, uses stats::quantile().

Usage

weighted_quantile(
  x,
  w = NULL,
  probs = seq(0, 1, 0.25),
  na.rm = TRUE,
  names = TRUE,
  ...
)

Arguments

x	Numeric vector.
w	Optional vector of non-negative case weights.
probs	Vector of probabilities.
na.rm	Ignore missing data? Default is TRUE.
names	Return names? Default is TRUE.
...	Further arguments passed to stats::quantile() in the unweighted case. Not used in the weighted case.

Value

A length-one numeric vector.

See Also

weighted_median()

Examples

n <- 10
x <- seq_len(n)
quantile(x)
weighted_quantile(x)
weighted_quantile(x, w = rep(1, n))
quantile(x, type = 1)
weighted_quantile(x, w = x) # same as Hmisc::wtd.quantile()
weighted_quantile(x, w = x, names = FALSE)
weighted_quantile(x, w = x, probs = 0.5, names = FALSE)

# Example with integer weights
x <- c(1, 1:11, 11, 11)
w <- seq_along(x)
weighted_quantile(x, w)
quantile(rep(x, w)) # same

MetricsWeighted documentation built on Nov. 16, 2023, 5:09 p.m.