ExtDist (version 0.6-3)

Gamma: The Gamma Distribution.

Description

Density, distribution, quantile, random number generation, and parameter estimation functions for the gamma distribution with parameters shape and scale. Parameter estimation can be based on a weighted or unweighted i.i.d sample and can be carried out numerically.

Usage

dGamma(x, shape = 2, scale = 2, params = list(shape = 2, scale = 2), ...)

pGamma(q, shape = 2, scale = 2, params = list(shape = 2, scale = 2), ...)

qGamma(p, shape = 2, scale = 2, params = list(shape = 2, scale = 2), ...)

rGamma(n, shape = 2, scale = 2, params = list(shape = 2, scale = 2), ...)

eGamma(X, w, method = c("moments", "numerical.MLE"), ...)

lGamma(X, w, shape = 2, scale = 2, params = list(shape = 2, scale = 2),
  logL = TRUE, ...)

Arguments

x,q

A vector of quantiles.

shape

Shape parameter.

scale

Scale parameter.

params

A list that includes all named parameters

...

Additional parameters.

p

A vector of probabilities.

n

Number of observations.

X

Sample observations.

w

An optional vector of sample weights.

method

Parameter estimation method.

logL

logical; if TRUE, lBeta_ab gives the log-likelihood, otherwise the likelihood is given.

Value

• dGamma gives the density, pGamma the distribution function, qGamma the quantile function, rGamma generates random deviates, and eGamma estimates the distribution parameters.lgamma provides the log-likelihood function.

Details

The dGamma(), pGamma(), qGamma(),and rGamma() functions serve as wrappers of the standard dgamma, pgamma, qgamma, and rgamma functions in the stats package. They allow for the parameters to be declared not only as individual numerical values, but also as a list so parameter estimation can be carried out. The gamma distribution with parameter shape=$\alpha$ and scale=$\beta$ has probability density function, $$f(x)= (1/\beta^\alpha \Gamma(\alpha))x^{\alpha-1}e^{-x/\beta}$$ where $\alpha > 0$ and $\beta > 0$. Parameter estimation can be performed using the method of moments as given by Johnson et.al (pp.356-357). The log-likelihood function of the gamma distribution is given by, $$l(\alpha, \beta |x) = (\alpha -1) \sum_i ln(x_i) - \sum_i(x_i/\beta) -n\alpha ln(\beta) + n ln \Gamma(\alpha)$$ where $\Gamma$ is the gamma function. The score function is provided by Rice (2007), p.270.

References

Johnson, N. L., Kotz, S. and Balakrishnan, N. (1995) Continuous Univariate Distributions, volume 1, chapter 17, Wiley, New York. Bury, K. (1999) Statistical Distributions in Engineering, Chapter 13, pp.225-226, Cambridge University Press. Rice, J.A. (2007) Mathematical Statistics and Data Analysis, 3rd Ed, Brookes/Cole.

See Also

ExtDist for other standard distributions.

Examples

# Parameter estimation for a distribution with known shape parameters
X <- rGamma(n=500, shape=1.5, scale=0.5)
est.par <- eGamma(X); est.par
plot(est.par)

#  Fitted density curve and histogram
den.x <- seq(min(X),max(X),length=100)
den.y <- dGamma(den.x,shape=est.par$shape,scale=est.par$scale)
hist(X, breaks=10, probability=TRUE, ylim = c(0,1.1*max(den.y)))
lines(den.x, den.y, col="blue")
lines(density(X), lty=2)

# Extracting shape or scale parameters
est.par[attributes(est.par)$par.type=="shape"]
est.par[attributes(est.par)$par.type=="scale"]

# Parameter estimation for a distribution with unknown shape parameters
# Example from:  Bury(1999) pp.225-226, parameter estimates as given by Bury are
# shape = 6.40 and scale=2.54. The log-likelihood for this data given
# Bury's parameter estimates is -656.7921.
data <- c(16, 11.6, 19.9, 18.6, 18, 13.1, 29.1, 10.3, 12.2, 15.6, 12.7, 13.1,
         19.2, 19.5, 23, 6.7, 7.1, 14.3, 20.6, 25.6, 8.2, 34.4, 16.1, 10.2, 12.3)
est.par <- eGamma(data, method="numerical.MLE"); est.par
plot(est.par)

# Estimates calculated by eGamma differ from those given by Bury(1999).
# However, eGamma's parameter estimates appear to be an improvement, due to a larger
# log-likelihood of -80.68186 (as given by lGamma below).

# log-likelihood
lGamma(data,param = est.par)

# Evaluating the precision of the parameter estimates by the Hessian matrix
H <- attributes(est.par)$nll.hessian
var <- solve(H)
se <- sqrt(diag(var));se