Group Regression Analysis and Visualization

In bregr: Easy and Efficient Batch Processing of Regression Models

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  out.width = "100%"
)

We are using the same example from the ezcox package vignette.

Load the package and data.

library(bregr)

data <- survival::lung
data <- data |>
  dplyr::mutate(
    ph.ecog = factor(ph.ecog),
    sex = ifelse(sex == 1, "Male", "Female")
  )

Construct grouped batch survival models to determine if the variable ph.ecog has different survival effects under different sex groups.

mds <- br_pipeline(
  data,
  y = c("time", "status"),
  x = "ph.ecog",
  group_by = "sex",
  method = "coxph"
)

We can examine the constructed models.

br_get_models(mds)

Now, display the results using a forest plot.

br_show_forest(mds)

We can optimize the plot for better visualization, for example, by removing the second column of the table and eliminating the row with NA results.

br_show_forest(
  mds,
  drop = 2,
  subset = !(Group_variable == "2" & variable == "ph.ecog" & label == 3)
)

To subset the data rows, we can input an R expression using variables from br_get_results(mds). For example, we can use Group_variable == "Female" & variable == "ph.ecog" & label == 3 to locate the row we want to remove, and then use !() to select the negated rows.

If drop All group is necessary, update the subset with:

br_show_forest(
  mds,
  drop = 2,
  subset = !((Group_variable == "Female" & variable == "ph.ecog" & label == 3) |
    (Group_variable == "All"))
)

bregr documentation built on June 28, 2025, 5:06 p.m.