pwMoment {EnvStats} | R Documentation |

## Estimate Probability-Weighted Moments

### Description

Estimate the `1jk`

'th probability-weighted moment from a random sample,
where either `j = 0`

, `k = 0`

, or both.

### Usage

```
pwMoment(x, j = 0, k = 0, method = "unbiased",
plot.pos.cons = c(a = 0.35, b = 0), na.rm = FALSE)
```

### Arguments

`x` |
numeric vector of observations. |

`j` , `k` |
non-negative integers specifying the order of the moment. |

`method` |
character string specifying what method to use to compute the
probability-weighted moment. The possible values are |

`plot.pos.cons` |
numeric vector of length 2 specifying the constants used in the formula for the
plotting positions when |

`na.rm` |
logical scalar indicating whether to remove missing values from |

### Details

The definition of a probability-weighted moment, introduced by
Greenwood et al. (1979), is as follows. Let `X`

denote a random variable
with cdf `F`

, and let `x(p)`

denote the `p`

'th quantile of the
distribution. Then the `ijk`

'th probability-weighted moment is given by:

`M(i, j, k) = E[X^i F^j (1 - F)^k] = \int^1_0 [x(F)]^i F^j (1 - F)^k \, dF`

where `i`

, `j`

, and `k`

are real numbers. Note that if `i`

is a
nonnegative integer, then `M(i, 0, 0)`

is the conventional `i`

'th moment
about the origin.

Greenwood et al. (1979) state that in the special case where `i`

, `j`

, and
`k`

are nonnegative integers:

`M(i, j, k) = B(j + 1, k + 1) E[X^i_{j+1, j+k+1}]`

where `B(a, b)`

denotes the beta function evaluated at
`a`

and `b`

, and

`E[X^i_{j+1, j+k+1}]`

denotes the `i`

'th moment about the origin of the `(j + 1)`

'th order
statistic for a sample of size `(j + k + 1)`

. In particular,

`M(1, 0, k) = \frac{1}{k+1} E[X_{1, k+1}]`

`M(1, j, 0) = \frac{1}{j+1} E[X_{j+1, j+1}]`

where

`E[X_{1, k+1}]`

denotes the expected value of the first order statistic (i.e., the minimum) in a
sample of size `(k + 1)`

, and

`E[X_{j+1, j+1}]`

denotes the expected value of the `(j+1)`

'th order statistic (i.e., the maximum)
in a sample of size `(j+1)`

.

*Unbiased Estimators* (`method="unbiased"`

)

Landwehr et al. (1979) show that, given a random sample of `n`

values from
some arbitrary distribution, an unbiased, distribution-free, and parameter-free
estimator of `M(1, 0, k)`

is given by:

`\hat{M}(1, 0, k) = \frac{1}{n} \sum^{n-k}_{i=1} x_{i,n} \frac{{n-i \choose k}}{{n-1 \choose k}}`

where the quantity `x_{i,n}`

denotes the `i`

'th order statistic in the
random sample of size `n`

. Hosking et al. (1985) note that this estimator is
closely related to U-statistics (Hoeffding, 1948; Lehmann, 1975, pp. 362-371).
Hosking et al. (1985) note that an unbiased, distribution-free, and parameter-free
estimator of `M(1, j, 0)`

is given by:

`\hat{M}(1, j, 0) = \frac{1}{n} \sum^n_{i=j+1} x_{i,n} \frac{{i-1 \choose j}}{{n-1 \choose j}}`

*Plotting-Position Estimators* (`method="plotting.position"`

)

Hosking et al. (1985) propose alternative estimators of `M(1, 0, k)`

and
`M(1, j, 0)`

based on plotting positions:

`\hat{M}(1, 0, k) = \frac{1}{n} \sum^n_{i=1} (1 - p_{i,n})^k x_{i,n}`

`\hat{M}(1, j, 0) = \frac{1}{n} \sum^n_{i=1} p_{i,n}^j x_{i,n}`

where

`p_{i,n} = \hat{F}(x_{i,n})`

denotes the plotting position of the `i`

'th order statistic in the random
sample of size `n`

, that is, a distribution-free estimate of the cdf of
`X`

evaluated at the `i`

'th order statistic. Typically, plotting
positions have the form:

`p_{i,n} = \frac{i-a}{n+b}`

where `b > -a > -1`

. For this form of plotting position, the
plotting-position estimators are asymptotically equivalent to the U-statistic
estimators.

### Value

A numeric scalar–the value of the `1jk`

'th probability-weighted moment
as defined by Greenwood et al. (1979).

### Note

Greenwood et al. (1979) introduced the concept of probability-weighted moments
as a tool to derive estimates of distribution parameters for distributions that
can be (perhaps only be) expressed in inverse form. The term “inverse form”
simply means that instead of characterizing the distribution by the formula for
its cumulative distribution function (cdf), the distribution is characterized by
the formula for the `p`

'th quantile (`0 \le p \le 1`

).

For distributions that can only be expressed in inverse form, moment estimates of their parameters are not available, and maximum likelihood estimates are not easy to compute. Greenwood et al. (1979) show that in these cases, it is often possible to derive expressions for the distribution parameters in terms of probability-weighted moments. Thus, for these cases the distribution parameters can be estimated based on the sample probability-weighted moments, which are fairly easy to compute. Furthermore, for distributions whose parameters can be expressed as functions of conventional moments, the method of probability-weighted moments provides an alternative to method of moments and maximum likelihood estimators.

Landwehr et al. (1979) use the method of probability-weighted moments to estimate the parameters of the Type I Extreme Value (Gumbel) distribution.

Hosking et al. (1985) use the method of probability-weighted moments to estimate the parameters of the generalized extreme value distribution.

Hosking (1990) and Hosking and Wallis (1995) show the relationship between probabiity-weighted moments and L-moments.

Hosking and Wallis (1995) recommend using the unbiased estimators of probability-weighted moments for almost all applications.

### Author(s)

Steven P. Millard (EnvStats@ProbStatInfo.com)

### References

Greenwood, J.A., J.M. Landwehr, N.C. Matalas, and J.R. Wallis. (1979).
Probability Weighted Moments: Definition and Relation to Parameters of Several
Distributions Expressible in Inverse Form. *Water Resources Research*
**15**(5), 1049–1054.

Hoeffding, W. (1948). A Class of Statistics with Asymptotically Normal
Distribution. *Annals of Mathematical Statistics* **19**, 293–325.

Hosking, J.R.M. (1990). L-Moments: Analysis and Estimation of Distributions
Using Linear Combinations of Order Statistics. *Journal of the Royal
Statistical Society, Series B* **52**(1), 105–124.

Hosking, J.R.M., and J.R. Wallis (1995). A Comparison of Unbiased and
Plotting-Position Estimators of L Moments. *Water Resources Research*
**31**(8), 2019–2025.

Hosking, J.R.M., J.R. Wallis, and E.F. Wood. (1985). Estimation of the
Generalized Extreme-Value Distribution by the Method of
Probability-Weighted Moments. *Technometrics* **27**(3), 251–261.

Landwehr, J.M., N.C. Matalas, and J.R. Wallis. (1979). Probability Weighted
Moments Compared With Some Traditional Techniques in Estimating Gumbel
Parameters and Quantiles. *Water Resources Research* **15**(5),
1055–1064.

Lehmann, E.L. (1975). *Nonparametrics: Statistical Methods Based on Ranks*.
Holden-Day, Oakland, CA, pp.362-371.

### See Also

### Examples

```
# Generate 20 observations from a generalized extreme value distribution
# with parameters location=10, scale=2, and shape=.25, then compute the
# 0'th, 1'st and 2'nd probability-weighted moments.
# (Note: the call to set.seed simply allows you to reproduce this example.)
set.seed(250)
dat <- rgevd(20, location = 10, scale = 2, shape = 0.25)
pwMoment(dat)
#[1] 10.59556
pwMoment(dat, 1)
#[1] 5.798481
pwMoment(dat, 2)
#[1] 4.060574
pwMoment(dat, k = 1)
#[1] 4.797081
pwMoment(dat, k = 2)
#[1] 3.059173
pwMoment(dat, 1, method = "plotting.position")
# [1] 5.852913
pwMoment(dat, 1, method = "plotting.position",
plot.pos = c(.325, 1))
#[1] 5.586817
#----------
# Clean Up
#---------
rm(dat)
```

*EnvStats*version 2.8.1 Index]