Gamma-Ordinal Link Function
Computes the gamma-ordinal transformation, including its inverse and the first two derivatives.
gordlink(theta, lambda = 1, cutpoint = NULL,
inverse = FALSE, deriv = 0, short = TRUE, tag = FALSE)theta |
Numeric or character. See below for further details. |
lambda, cutpoint |
The former is the shape parameter in |
inverse, deriv, short, tag |
Details at |
The gamma-ordinal link function (GOLF) can be applied to a parameter lying in the unit interval. Its purpose is to link cumulative probabilities associated with an ordinal response coming from an underlying 2-parameter gamma distribution.
See Links for general information about VGAM
link functions.
See Yee (2019) for details.
Numerical values of theta too close to 0 or 1 or out of range
result in large positive or negative values, or maybe 0 depending on
the arguments.
Although measures have been taken to handle cases where
theta is too close to 1 or 0,
numerical instabilities may still arise.
In terms of the threshold approach with cumulative probabilities for
an ordinal response this link function corresponds to the
gamma distribution (see gamma2) that has been
recorded as an ordinal response using known cutpoints.
Thomas W. Yee
Yee, T. W. (2020). Ordinal ordination with normalizing link functions for count data, (in preparation).
## Not run:
gordlink("p", lambda = 1, short = FALSE)
gordlink("p", lambda = 1, tag = TRUE)
p <- seq(0.02, 0.98, len = 201)
y <- gordlink(p, lambda = 1)
y. <- gordlink(p, lambda = 1, deriv = 1, inverse = TRUE)
max(abs(gordlink(y, lambda = 1, inverse = TRUE) - p)) # Should be 0
#\ dontrun{par(mfrow = c(2, 1), las = 1)
#plot(p, y, type = "l", col = "blue", main = "gordlink()")
#abline(h = 0, v = 0.5, col = "orange", lty = "dashed")
#plot(p, y., type = "l", col = "blue",
# main = "(Reciprocal of) first GOLF derivative")
#}
# Another example
gdata <- data.frame(x2 = sort(runif(nn <- 1000)))
gdata <- transform(gdata, x3 = runif(nn))
gdata <- transform(gdata, mymu = exp( 3 + 1 * x2 - 2 * x3))
lambda <- 4
gdata <- transform(gdata,
y1 = rgamma(nn, shape = lambda, scale = mymu / lambda))
cutpoints <- c(-Inf, 10, 20, Inf)
gdata <- transform(gdata, cuty = Cut(y1, breaks = cutpoints))
#\ dontrun{ par(mfrow = c(1, 1), las = 1)
#with(gdata, plot(x2, x3, col = cuty, pch = as.character(cuty))) }
with(gdata, table(cuty) / sum(table(cuty)))
fit <- vglm(cuty ~ x2 + x3, cumulative(multiple.responses = TRUE,
reverse = TRUE, parallel = FALSE ~ -1,
link = gordlink(cutpoint = cutpoints[2:3], lambda = lambda)),
data = gdata, trace = TRUE)
head(depvar(fit))
head(fitted(fit))
head(predict(fit))
coef(fit)
coef(fit, matrix = TRUE)
constraints(fit)
fit@misc
## End(Not run)Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.