bootSummary: 'bootSummary' calculates the empirical distribution of a...

In West-End-Statistics/r-library-vakdr: Miscellanious helper functions

Description

The bootSummary function uses bootstaps to estimate the midpoint, avg and confidence interval of a user defined function. It's purpose is to work seamlessly within the dplyr framework/the pipe an allows the use of bare column names.

Usage

bootSummary(data, var, ..., .funs = median, n = 100, ci = 0.95, na.rm = FALSE)

Arguments

data	data.frame or tibble.
var	bare column name to summarise over.
...	grouping variables for summary statistic.
.funs	summarising function. It can be a bare function name or follow the usage of funs.
n	Number of bootstrap replicates to generate
ci	width of quantile interval for final summary.
na.rm	should the final summarization across bootraps remove NAs?

Details

The user provides the name of column to summarise along with the summarise function.

The example shows how a t-test performs similarly to a bootstrap when the data is normal (Group A). It's also possible to make estimates for other statistitics such as the median.

Value

a tibble containing the name of the grouping variables and the following columns:

stat_mean

The mean across bootstraps

stat_mid

The median across bootstraps

stat_low

The low quantile (e.g. 2.5% when ci = .95)

stat_high

The high quantile (e.g. 97.5% when ci = .95)

Examples

library(dplyr)
# Simulate some data
set.seed(5)
size <- 1000
test_data <- data.frame(cohort = rep(c("A", "B"), each = size),
                        stat = c(rnorm(size, 5, 10), exp(rnorm(size, mean = 0.1))))
# T Tests
test_data %>%
  filter(cohort == "A") %>%
  pull(stat) %>% t.test()
test_data %>%
  filter(cohort == "B") %>%
  pull(stat) %>% t.test()

# Bootrap the median
test_data %>% bootSummary(stat, cohort)

West-End-Statistics/r-library-vakdr documentation built on Dec. 18, 2021, 7:16 p.m.