update.gkwreg: Update and Re-fit a GKw Regression Model

In gkwreg: Generalized Kumaraswamy Regression Models for Bounded Data

Update and Re-fit a GKw Regression Model

Description

Updates and (by default) re-fits a Generalized Kumaraswamy regression model. This method allows modification of the model formula, data, or other arguments without having to completely re-specify the model call. Supports formulas with up to 5 parts (alpha, beta, gamma, delta, lambda) using the Formula package.

Usage

## S3 method for class 'gkwreg'
update(object, formula., ..., data. = NULL, evaluate = TRUE)

Arguments

object	An object of class "gkwreg", typically obtained from gkwreg.
formula.	Changes to the formula. This is a formula where . refers to the corresponding part of the old formula. For multi-part formulas (e.g., y ~ x1 | x2 | x3), you can update each part separately using the | separator.
...	Additional arguments to the call, or arguments with changed values. Use name = NULL to remove an argument.
data.	Optional. A new data frame in which to evaluate the updated model. If omitted, the original data is used.
evaluate	Logical. If TRUE (default), the updated model is fitted. If FALSE, the updated call is returned without fitting.

Details

The update method allows you to modify a fitted model and re-fit it with the changes. The GKw regression model supports formulas with up to 5 parts: y ~ model_alpha | model_beta | model_gamma | model_delta | model_lambda

Each part can be updated independently using . to refer to the current specification:

• . ~ . + x | . | . | . | . - Add x to alpha only

• . ~ . | . + x | . | . | . - Add x to beta only

• . ~ . | . | . + x | . | . - Add x to gamma only

• . ~ . + x | . + x | . | . | . - Add x to alpha and beta

• . ~ . - x | . | . | . | . - Remove x from alpha

Omitting parts at the end is allowed (they default to .):

• . ~ . + x | . is equivalent to . ~ . + x | . | . | . | .

• . ~ . | . + x is equivalent to . ~ . | . + x | . | . | .

Value

If evaluate = TRUE, a new fitted model object of class "gkwreg". If evaluate = FALSE, an updated call.

Author(s)

Lopes, J. E.

See Also

gkwreg, update, Formula

Examples

# Load example data
require(gkwreg)

data(GasolineYield)

# EXAMPLE 1: Simple formulas (1 part - alpha only)

m1_0 <- gkwreg(yield ~ 1, data = GasolineYield, family = "kw")
m1_1 <- update(m1_0, . ~ . + temp)
m1_2 <- update(m1_1, . ~ . + batch)
m1_3 <- update(m1_2, . ~ . - temp)

anova(m1_0, m1_1, m1_2)
AIC(m1_0, m1_1, m1_2, m1_3)
BIC(m1_0, m1_1, m1_2, m1_3)

# EXAMPLE 2: Two-part formulas (alpha | beta)

# Start with intercept-only for both
m2_0 <- gkwreg(yield ~ 1 | 1, data = GasolineYield, family = "kw")

# Add temp to alpha
m2_1 <- update(m2_0, . ~ . + temp | .)

# Add batch to beta
m2_2 <- update(m2_1, . ~ . | . + batch)

# Add batch to alpha too
m2_3 <- update(m2_2, . ~ . + batch | .)

anova(m2_0, m2_1, m2_2, m2_3)
AIC(m2_0, m2_1, m2_2, m2_3)

# EXAMPLE 3: Three-part formulas (alpha | beta | gamma)

m3_0 <- gkwreg(yield ~ 1,
  data = GasolineYield,
  family = "gkw",
  control = gkw_control(method = "BFGS", maxit = 2000)
)

m3_1 <- update(m3_0, . ~ . + temp | . | .)
m3_2 <- update(m3_1, . ~ . | . + batch | .)
m3_3 <- update(m3_2, . ~ . | . | . + temp)

anova(m3_0, m3_1, m3_2, m3_3)

# EXAMPLE 4: Practical nested model comparison

# Null model
fit0 <- gkwreg(yield ~ 1,
  data = GasolineYield,
  family = "kw",
  control = gkw_control(method = "BFGS", maxit = 2000)
)

# Add main effects to alpha
fit1 <- update(fit0, . ~ . + temp)
fit2 <- update(fit1, . ~ . + batch)

# Model beta parameter
fit3 <- update(fit2, . ~ . | temp)
fit4 <- update(fit3, . ~ . | . + batch)

# Full comparison
anova(fit0, fit1, fit2, fit3, fit4)
AIC(fit0, fit1, fit2, fit3, fit4)
BIC(fit0, fit1, fit2, fit3, fit4)

# EXAMPLE 5: Changing other parameters

# Change family
fit_gkw <- update(fit2, family = "gkw")

# Change link function
fit_logit <- update(fit2, link = list(alpha = "logit"))

# View call without fitting
update(fit2, . ~ . | . + temp, evaluate = FALSE)

gkwreg documentation built on Nov. 27, 2025, 5:06 p.m.