tableRegression: Format coefficient tables of regression models

In felix-hof/biostatUZH: Misc Tools of the Department of Biostatistics, EBPI, University of Zurich

Format coefficient tables of regression models

Description

Formats output from the regression model functions: lm, glm, glm.nb, coxph, and Weibull.

Usage

tableRegression(
  model,
  stats = NULL,
  col.names = NULL,
  row.names = NULL,
  intercept = NULL,
  text = c("english", "german"),
  text.ci = text,
  eps.pvalue = 1e-04,
  digits = NULL,
  strict = FALSE,
  big.mark = "'",
  xtable = TRUE,
  align = NULL,
  caption = NULL,
  label = NULL,
  vars = NULL,
  ...
)

Arguments

model	Object of class lm, glm, negbin (obtained by glm.nb), coxph (obtained by coxph), and list obtained by Weibull.
stats	character vector with stats chosen from "estimate", "exp.estimate", "standarderror", "t.value", "ci.95", and "p.value".
col.names	Character vector of same length and order as in stats. A percentage sign must be escaped by two backslashes.
row.names	Character vector of row names.
intercept	Logical vector of length one indicating whether to provide an intercept or not. If intercept is set TRUE, the first line of the summary output is removed. If the model is a binomial regression, intercept is set FALSE. Intercepts are not available for Weibull or Cox models, because they do not provide any intercept value.
text	Either "english" (default) or "german" indicating the used language names.
text.ci	Either "english", "german" or "none". The language used to denote confidence interval, see formatCI.
eps.pvalue	If strict = FALSE, p-values smaller than eps.pvalue will be formatted as "< eps.pvalue". Otherwise this argument is ignored.
digits	Vector of length stats, digits used for each column.
strict	Either TRUE or FALSE (default). If FALSE, p-values are formatted with formatPval. Otherwise formatPvalStrict is used to format p-values. This argument is thus only relevant when the argument stats is either NULL or contains "p.value".
big.mark	Character vector as in format.
xtable	If TRUE, a Latex table is returned, otherwise a data.frame is returned.
align	See xtable.
caption	See xtable.
label	See xtable.
vars	Specify the variables for which regression summaries should be printed. The argument vars takes a string vector with the names of the coefficients in the model.
...	Arguments passed to print.xtable.

Details

In stats:

• If t.value is chosen, the z.value might be taken, depending on the model.

• For lm-models: ci.95 calculates a confidence interval for the estimate.

• For glm- and coxph-models: ci.95 calculates a confidence interval for the exp(estimate).

Value

Depending on the value of the xtable argument, the function either prints and returns LaTeX code representing the produced table of coefficients, or it returns the corresponding data frame.

Author(s)

Sina Rueeger with contributions by Sebastian Meyer.

See Also

tableOR, xtable, lm, glm, glm.nb coxph, Weibull.

Examples

## Linear model
## ---------------
mod.lm <- lm(Sepal.Length ~ Sepal.Width, data = iris)
mod.lm1 <- lm(Sepal.Length ~ .^2, data = iris) 

tableRegression(model = mod.lm)

## strict argument is used to force p-values to have a certain number
## of digits (here 4).
tableRegression(model = mod.lm, digits = c(2, 2, 4), strict = TRUE)
tableRegression(model = mod.lm, digits = c(2, 2, 4), strict = FALSE)

## choosing columns, columns and row naming in german
tableRegression(model = mod.lm1, stats = c("estimate", "t.value", "p.value"),
                text = "german")

## adapt row names, plus special format for ci
tableRegression(model = mod.lm, row.names = c("Intercept", "Width Sepal"),
                text.ci = "none")

## Poisson model
## (example from ?glm)
## --------------
counts <- c(18,17,15,20,10,20,25,13,12)
outcome <- gl(3,1,9)
treatment <- gl(3,3)
d.AD <- data.frame(treatment, outcome, counts)
mod.glm.pois <- glm(counts ~ outcome + treatment, family=poisson())
tableRegression(model = mod.glm.pois)


## Negative binomial model
## --------------
mod.glm.nb <- glm.nb(Days ~ Sex + Age, data = quine)
tableRegression(model = mod.glm.nb,
    caption = paste("NegBin model. Estimated dispersion:",
        sprintf("%4.2f ($se=%4.2f$).", mod.glm.nb$theta, mod.glm.nb$SE.theta)),
    label = "tab:glm.nb")



## Logistic model
## -------------
dat <- survival::rats
dat$rx <- factor(dat$rx, labels = c(" (A)", " (B)"))
mod.glm.bin <- glm(status ~ litter + rx, family = binomial, data = dat)

tableRegression(model = mod.glm.bin,
                stats = c("estimate", "exp.estimate", "ci.95", "t.value", "p.value"),
                text = "english", digits = rep(3, 5),
                caption = "Here goes the caption.", label = "mod:logit")

## including intercept
tableRegression(model = mod.glm.bin,
                stats = c("estimate", "exp.estimate", "ci.95", "t.value", "p.value"),
                text = "english", digits = rep(3, 5),
                caption = "Here goes the caption.", label = "mod:logit",
                intercept = TRUE)


## Cox model
## (example from ?survival::coxph)
## -------------
dat <- list(time = c(4, 3, 1, 1, 2, 2, 3), 
            status = c(1, 1, 1, 0, 1, 1, 0), 
            x = c(0, 2, 1, 1, 1, 0, 0), 
            sex = c(0, 0, 0, 0, 1, 1, 1)) 

mod.cox <- coxph(Surv(time, status) ~ x, data = dat)
mod.cox1 <- coxph(Surv(time, status) ~ x + factor(sex), data = dat)
mod.cox2 <- coxph(Surv(time, status) ~ x + strata(sex), data = dat)

tableRegression(model = mod.cox)
tableRegression(model = mod.cox1)
tableRegression(model = mod.cox2)


## Weibull
## (example from WeibullReg)
## -------------
data("larynx")
mod.wb <- weibullReg(Surv(time, death) ~ factor(stage) + age,
                     data = larynx)
tableRegression(model = mod.wb)

felix-hof/biostatUZH documentation built on Sept. 27, 2024, 1:48 p.m.