binomial_glm_plot: Plot for a given logistic regression model (glm with binomial...

In model4you: Stratified and Personalised Models Based on Model-Based Trees and Forests

Description

Can be used on its own but is also useable as plotfun in node_pmterminal.

Usage

binomial_glm_plot(
  mod,
  data = NULL,
  plot_data = FALSE,
  theme = theme_classic(),
  ...
)

Arguments

mod	A model of class glm with binomial family.
data	optional data frame. If NULL the data stored in mod is used.
plot_data	should the data in form of a mosaic type plot be plotted?
theme	A ggplot2 theme.
...	ignored at the moment.

Examples

set.seed(2017)

# number of observations
n <- 1000

# balanced binary treatment
# trt <- factor(rep(c("C", "A"), each = n/2),
#               levels = c("C", "A"))

# unbalanced binary treatment
trt <- factor(c(rep("C", n/4), rep("A", 3*n/4)),
              levels = c("C", "A"))

# some continuous variables
x1 <- rnorm(n)
x2 <- rnorm(n)

# linear predictor
lp <- -0.5 + 0.5*I(trt == "A") + 1*I(trt == "A")*I(x1 > 0)

# compute probability with inverse logit function
invlogit <- function(x) 1/(1 + exp(-x))
pr <- invlogit(lp)

# bernoulli response variable
y <- rbinom(n, 1, pr)
dat <- data.frame(y, trt, x1, x2)

# logistic regression model
mod <- glm(y ~ trt, data = dat, family = "binomial")
binomial_glm_plot(mod, plot_data = TRUE)

# logistic regression model tree
ltr <- pmtree(mod)
plot(ltr, terminal_panel = node_pmterminal(ltr,
                                           plotfun = binomial_glm_plot,
                                           confint = TRUE,
                                           plot_data = TRUE))

model4you documentation built on Jan. 8, 2021, 3:02 p.m.