logmean: Log-Average or Geometric Mean

In robertdouglasmorrison/DuffyTools: Duffy Lab Utility Tools

Log-Average or Geometric Mean

Description

Compute the log-average or geometric mean of a numeric vector

Usage

logmean(x, na.rm = FALSE)

logMeanPlusSD(x, na.rm = FALSE)
logMedianPlusSD(x, na.rm = FALSE)

Arguments

x	numeric vector of positive values
nSD	numeric scalar, the number of standard deviations to add

Details

Implements the geometric mean, by log transforming the values in x, calculating that mean, and then exponentiating the result. This has better numerical stability then then using the N-th root of the power series of x.

The 'PlusSD' forms are used to thresholds of statistical significance, by adding an arbitrary number of standard deviation of log(x), to the calculation.

As log transform is undefined for zeros and negative values, some level of data cleaning is implemented to tolerate invalid data. Specifically, negatives are set to NA and thus ignored. If all values are zero, zero is returned. If a subset of values are zero, they are replaced with the smallest non-zero value prior to log transform and averaging.

Value

a numeric scalar

See Also

Contrast with mean and sqrtmean.

robertdouglasmorrison/DuffyTools documentation built on April 16, 2024, 6:31 a.m.