gm_conf: Calculate the geometric X percent confidence interval

View source: R/gm_mean.R

gm_confR Documentation

Calculate the geometric X percent confidence interval

Description

gm_conf takes as input a numeric vector, the desired confidence interval, and returns the lower and upper values for the confidence interval. Note that, because calculating a confidence interval relies on an accurate estimate of the mean and standard deviation, your data should be log-normally distributed.

Usage

gm_conf(
  x,
  CI = 0.9,
  na.rm = TRUE,
  zero.propagate = FALSE,
  distribution_type = "t"
)

Arguments

x

A vector of numbers

CI

The confidence interval desired; default is 90 as a decimal, e.g., 0.95.

na.rm

Should NA values be removed? (logical)

zero.propagate

Should zeroes be propagated? (logical)

distribution_type

use a "t" distribution (default) or a "Z" distribution. Note: The Simcyp Simulator calculates geometric confidence intervals with a t distribution.

Examples

x <- rnorm(100, 5, 1)
gm_conf(x)
gm_conf(x, CI = 0.9)

# PK data are often log-normally distributed, so try this function with
# some example concentration-time data. (For this, we're not worried
# about independence; we just need some example data to work with.)
data(MDZConcTime)
# Making values larger just so we're not dealing w/tiny decimals and
# removing 0's for simplicity.
x <- MDZConcTime$Conc[MDZConcTime$Conc != 0]*100

# Compare the distributions of the untransformed vs. log-transformed data:
ggplot2::qplot(x, bins = 15)
ggplot2::qplot(x, bins = 15) + ggplot2::scale_x_log10()
gm_conf(x)
# Compare the results with confInt()
confInt(x)


shirewoman2/Consultancy documentation built on Feb. 18, 2025, 10 p.m.