my.dsn {RFlocalfdr}

my.dsn

Description

density of skew-normal using the appromimation of Ashour, Samir K. and Abdel-hameed, Mahmood A.

Usage

my.dsn(x, xi = 0, omega = 1, lambda = 1)

dsn(x, xi = 0, omega = 1, alpha = 0, tau = 0)

psn(q, xi = -Inf, omega = 1, alpha = 0, tau = 0, ...)

qsn(p, xi = Inf, omega = 1, alpha = 0, tau = 0, ...)

Arguments

x

vector of quantiles. Missing values (‘NA’'s) and ‘Inf’'s are allowed.

xi

vector of location parameters.

omega

vector of scale parameters; must be positive

lambda

param

alpha

vector of slant parameter(s)

tau

= 0

q

vector of quantiles

...

arguments passed to sn

p

vector of probabilities. Missing values (‘NA’'s) are allowed

Details

See https://en.wikipedia.org/wiki/Skew_normal_distribution for discussion of the skew-normal. Using the appromimation of Ashour, Samir K. and Abdel-hameed, Mahmood A. "Approximate Skew Normal Distribution", Journal of Advanced Research, 2010, 1:4. It accepts the parameters xi, omega, lambda (Ashour et. al. 2010). Other foumulations may use different parameterizations. The sn (skew-normal) package incluse the extended skew-normal (ESN) distribution. For the SN the tau parameter is 0.

RFlocalfdr also uses wrappers around the functions dsn, qsn and psn from the package "sn" https://cran.r-project.org/web/packages/sn/. This is due to the fact that fitdistrplus::fitdist(imp, "sn", start = list(xi = mean(imp)... returns warnings such as The dsn function should return a zero-length vector when input has length zero and not raise an error The psn function should have its first argument named: q as in base R These wrappers ensure conformity with the expectations of fitdistrplus::fitdist

Value

fits the Density function for the skew-normal (SN) distribution.

Examples

library(sn)
curve(sn::dsn(x,xi=0, omega=1, alpha=1, tau=0),xlim=c(-10,10),col="blue")
curve(sn::dsn(x,xi=0, omega=1, alpha=0.1, tau=0),xlim=c(-10,10),col="blue",add=TRUE)
curve(sn::dsn(x,xi=1, omega=2, alpha=2, tau=0),xlim=c(-10,20),col="blue",add=TRUE)
curve(sn::dsn(x,xi=3, omega=4, alpha=4, tau=0),xlim=c(-10,20),col="blue",add=TRUE)

curve(my.dsn(x),xlim=c(-10,10),col="red",add=TRUE)
curve(my.dsn(x,lambda=0.1),xlim=c(-10,10),col="red",add=TRUE)
curve(my.dsn(x,xi=1, omega=2, lambda=2),xlim=c(-10,20),col="red",add=TRUE)
curve(my.dsn(x,xi=3, omega=4, lambda=4),xlim=c(-10,20),col="red",add=TRUE)

#dsn, qsn and psn are wrappers around the provided functions provided by sn. This is done to
# overcome some checking done by fitdistrplus

library(sn)
getAnywhere("dsn")
RFlocalfdr::my.test1fun("sn::dsn", list(xi = -Inf, omega =1, alpha=0 ), fix.arg = list(tau = 0))
RFlocalfdr::my.test1fun("sn::psn", list(xi = -Inf, omega =1, alpha=0 ), fix.arg = list(tau = 0))
RFlocalfdr::my.test1fun("sn::qsn", list(xi = -Inf, omega =1, alpha=0 ), fix.arg = list(tau = 0))
#all return FALSE

detach("package:sn", unload=TRUE)
getAnywhere("dsn")
RFlocalfdr::my.test1fun("dsn", list(xi = -Inf, omega =1, alpha=0 ), fix.arg = list(tau = 0))#TRUE
RFlocalfdr::my.test1fun("psn", list(xi = -Inf, omega =1, alpha=0 ), fix.arg = list(tau = 0))#TRUE
RFlocalfdr::my.test1fun("qsn", list(xi = -Inf, omega =1, alpha=0 ), fix.arg = list(tau = 0))#TRUE