count_sigfigs: Count number significant figures

In slin30/wzMisc: Miscellaneous functions by WZ

Count number significant figures

Description

Count number of significant digits (figures), option to include trailing zeros

Usage

count_sigfigs(x, digits = getOption("digits"), countTrailing = FALSE)

Arguments

x	numeric or character vector; only character input is allowed if countTrailing=TRUE.
digits	integer of length 1. Optional. Default uses value from getOptions("digits"). Will warn if this is set higher than 15, for good reason. See notes.
countTrailing	logical. Should trailing zeros be counted? Defaults to FALSE.

Details

Count number of significant digits, either from a (measurement) precision or a (mathematical) numeric perspective via the countTrailing option. When set to TRUE and provided a character input x, this option will count trailing zeros as significant digits. Leading zeros are (almost) never counted; the single exception is when countTrailing=TRUE and x contains values that would be coerced to 0 via e.g. as.integer or as.numeric; see Notes.

By default, countTrailing=FALSE; this counts significant digits from a (mathematical) numeric perspective. Setting to TRUE counts significant digits from a (measurement) precision perspective.

Because R (and most other analysis tools/languages) will automatically strip trailing zeros for integer or numeric data types, this function will throw an error if countTrailing=TRUE and class(x) != "character".

This function does not perform any explicit rounding or truncation of input within reasonable limits; if input precision exceeds system limit, standard R rounding rules will apply, silently.

Value

An integer vector of significant digit counts equal in length to input vector.

Note

Zero is a special case. If countTrailing=FALSE, will return zero. If countTrailing=TRUE, all digits are counted, excluding digits following the exponent notation (e or E) if applicable. See examples

This function assumes that if you provide character inputs that would evaluate to 0 if treated as a number, this is in fact meaningful from a precision/measurement perspective.

Examples

set.seed(1)
x <- mapply(signif, rnorm(10), 1:10)
count_sigfigs(x)
count_sigfigs(x, digits = 10)

#Also works for scientific notation
count_sigfigs(c(1E10, -1E10, -000001E-10)) #all 1
count_sigfigs(c("1.0001", 1.0001, 0001.0001)) #all 5

#digits option
#By default, will use local number of digits, which is 7
getOption("digits") #default should be 7
count_sigfigs(0.1234567) #7
count_sigfigs(0.12345678) #8
count_sigfigs(0.123456789) #8
count_sigfigs(0.123456789, digits = 8) #9

#countTrailing
count_sigfigs("1.0100", countTrailing = TRUE)
count_sigfigs("1.0100E10", countTrailing = TRUE)

#zero handling
zeros_decimal  <- c("0", "0.0", "00.0", "00.00", "0.000")
zeros_exponent <- c("0E0", "0.0E0", "00.0E0", "00.00E0", "0.000E0")
count_sigfigs(zeros_decimal) #all zero
count_sigfigs(zeros_decimal, countTrailing = TRUE) #1,2,3,4,4
count_sigfigs(zeros_exponent) #all zero
count_sigfigs(zeros_exponent, countTrailing = TRUE) #also 1,2,3,4,4

## Not run: 
#compare:
count_sigfigs(123.12345679, digits = 7)  #need higher digits option
count_sigfigs(123.12345679, digits = 15) #what you expected
count_sigfigs(123.12345679, digits = 16) #probably not what you expected

#also
count_sigfigs(1000000001.12345, digits = 15) #pushing it

## End(Not run)

slin30/wzMisc documentation built on Jan. 27, 2023, 1 a.m.