est_param_2: Estimation of mean and/or variance for models fitted with...

In fhernanb/model: This Package Contains Useful Functions for Modeling Regresion

Estimation of mean and/or variance for models fitted with gamlss2.

Description

This function estimates parameters like mean, standard deviation for a gamlss2 object. This function is very useful when the mean or standard deviation do not match with the parameters \mu, \sigma, \nu and \tau, ..., or some function of the parameters \mu, \sigma, \nu and \tau.

Usage

est_param_2(mod, fun = "mean", m = 100, data, newdata = NULL, ...)

Arguments

mod	a gamlss2 object.
fun	the estimated parameter, by default is mean but can be changed for sd, var or any statistical function.
m	a number to indicate the number of observations to simulate, by default its values is 100.
data	the original data frame used to fit the mod.
newdata	a data frame containing new values for the explanatory variables.
...	additional parameters for "fun" function.

Details

The function obtains the fitted values for mu, sigma, nu and tau. The functions simulates m observations and then it obtains the mean/fun defined by the user.

Value

est_param_2 function returns a vector.

Examples

# Example for NOrmal regression ------------------------
# Here E(y) = mu and Var(y)= sigma
n <- 7
x <- runif(n=n)
y <- rnorm(n=n, mean=-5 + 10 * x, sd=exp(1 + 2 * x))
data1 <- data.frame(y, x)
data1

library(gamlss2)
mod1 <- gamlss2(y ~ x, sigma.fo=~x, family=NO, data=data1)

# Obtaining the means
m1 <- fitted(mod1, model="mu", type="parameter")
m2 <- est_param_2(mod1, m=10000, fun="mean")
cbind(m1, m2) # Comparing the estimated means
cor(m1, m2)

# Obtaining the standard deviations
s1 <- fitted(mod1, model="sigma", type="parameter")
s2 <- est_param_2(mod1, m=10000, fun="sd")
cbind(s1, s2) # Comparing the estimated standard deviations
cor(s1, s2)

# Example for GAmma regression --------------------------
# Here E(y) = mu and Sd(y) = mu * sigma
n <- 7
x <- runif(n=n)
y <- rGA(n=n, mu=exp(-1 + 5 * x), sigma=exp(-1 + 2 * x))
data2 <- data.frame(y, x)
data2

mod2 <- gamlss2(y ~ x, sigma.fo=~x, family=GA, data=data2)

summary(mod2)

# Obtaining the means
m1 <- fitted(mod2, model="mu", type="parameter")
m2 <- est_param_2(mod2, m=10000, fun="mean")
cbind(m1, m2) # Comparing the estimated means
cor(m1, m2)

# Obtaining the standard deviations
s1 <- fitted(mod2, what="sigma") * fitted(mod2, what="mu")
s2 <- est_param_2(mod2, m=10000, fun="sd")
cbind(s1, s2) # Comparing the estimated standard deviations
cor(s1, s2)

# To predict mean and standard deviation for new observations
new_data <- data.frame(x=c(0.26, 0.52, 0.77))

est_param_2(mod=mod2, fun="mean", m=10000, data=data2, newdata=new_data)
est_param_2(mod=mod2, fun="sd"  , m=10000, data=data2, newdata=new_data)

fhernanb/model documentation built on June 12, 2025, 3:12 p.m.