normal.bound: Normal Approximation of the Maximum Error

In koenderks/auditR: A Multifunctional R Package for Auditing

Description

Calculates a confidence bound for the maximum error in an audit population according to the Normal distribution.

Usage

normal.bound(bookValues, auditValues, confidence = 0.95)

Arguments

bookValues	A vector of book values from sample.
auditValues	A vector of corresponding audit values from the sample.
confidence	The amount of confidence desired from the bound (on a scale from 0 to 1), defaults to 95% confidence.

Value

An estimate of the mean taint per dollar unit in the population.

Details

EMPTY FOR NOW

Author(s)

Koen Derks, [e-mail]

Examples

# Create an imaginary data set
bookValues   <- rgamma(n = 2400, shape = 1, rate = 0.001)
error.rate   <- 0.1
error        <- sample(0:1, 2400, TRUE, c(1-error.rate, error.rate))
taint        <- rchisq(n = 2400, df = 1) / 10
auditValues  <- bookValues - (error * taint * bookValues)
frame        <- data.frame( bookValues = round(bookValues,2),
                            auditValues = round(auditValues,2))
# Draw a sample
samp.probs   <- frame$bookValues/sum(frame$bookValues)
sample.no    <- sample(1:nrow(frame), 100, FALSE, samp.probs)
sample       <- frame[sample.no, ]
# Calculate bound
normal.bound(bookValues = sample$bookValues,
             auditValues = sample$auditValues,
             confidence = 0.95)

koenderks/auditR documentation built on May 16, 2019, 7:16 p.m.