valreg: Function to validate a regression model

In Antoine-Masse/KefiR: KefiR Package

Function to validate a regression model

Description

Function to validate a regression model

Usage

valreg(
  reg,
  verbose = TRUE,
  nvar = 5,
  boot = TRUE,
  alpha = 0.05,
  conf.level = 0.95,
  plot = FALSE,
  data = c(),
  raintest_alpha = 0.05,
  dwtest_alpha = 0.03,
  shapiro_alpha = 0.05,
  bptest_alpha = 0.05
)

Arguments

reg	A regression model
verbose	To see the detailed balance sheet
nvar	The maximum number of variables allowed
boot	For checking the model by bootstrap with bootreg() (FALSE or TRUE)
alpha	Maximum value accepted for the p-values of the model and its coefficients
conf.level	Confidence interval accepted to validate the regression model by bootstrap
plot	For seeing the graphical analysis of bootreg() (FALSE or TRUE)
data	optional, the data.frame of data if complex model.
raintest_alpha	Minimal value of p-value accepted for Rainbow test
dwtest_alpha	Minimal value of p-value accepted for Durbin-Watson test
shapiro_alpha	Minimal value of p-value accepted for Shapiro-Wilk test
bptest_alpha	Minimal value of p-value accepted for Breush-Pagan test

Value

This function allows to run all the tests necessary to validate a regression model (check the normal distribution of the residuals, avoid leverage effects, control the variance of the residuals...).

valreg will therefore validate the regression model, control the p-values and, possibly (boot argument), control the reliability by bootstrap with the bootreg function.

Author(s)

Antoine MASSE (2024)

Examples

# Example 1: Linear model
data(iris)
reg <- lm(Sepal.Length~.,data=iris[,1:4])
valreg(reg,verbose=TRUE)

# Example 2: Mixed model
library(lme4)
data(sleepstudy)
reg_mixed <- lmer(Reaction ~ Days + (Days | Subject), data = sleepstudy)
valreg(reg_mixed, verbose=TRUE)

# Example 3: Mixed model
library(lme4)
data(iris)
reg_mixed <- lmer(Sepal.Length ~ Petal.Length + (1|Species), data = iris)
valreg(reg_mixed, verbose=TRUE)

Antoine-Masse/KefiR documentation built on July 4, 2024, 11:40 a.m.