A simple index of "information" content associated with individuals in a SoilProfileCollection object. Information content is quantified by number of bytes after gzip compression via memCompress().

profileInformationIndex(
  x,
  vars,
  method = c("median", "mean", "sum"),
  baseline = TRUE,
  useDepths = TRUE,
  numericDigits = 4
)

Arguments

x

SoilProfileCollection object

vars

character vector of site or horizon level attributes to consider

method

character: aggregation method, information content evaluated over vars: 'median', 'mean', or 'sum'

baseline

logical, compute ratio to "baseline" information content, see details

useDepths

logical, include horizon depths in vars

numericDigits

integer, number of significant digits to retain in numeric -> character conversion

Value

a numeric vector of the same length as length(x) and in the same order, suitable for direct assignment to a new site-level attribute

Details

Information content via compression (gzip) is the central assumption behind this function: the values associated with a simple soil profile having few horizons and little variation between horizons (isotropic depth-functions) will compress to a much smaller size than a complex profile (many horizons, strong anisotropy). Information content is evaluated a profile at a time, over each site or horizon level attribute specified in vars. Values are aggregated to the profile level by method: median, mean, or sum. The baseline argument invokes a comparison to the simplest possible representation of each depth-function:

  • numeric: replication of the mean value to match the number of horizons with non-NA values

  • character or factor: replication of the most frequent value to match the number of horizons with non-NA values

The ratios computed against a "simple" baseline represent something like "information gain", ranging from 0 to 1. Larger baseline ratios suggest more complexity (more information) associated with a soil profile's depth-functions. Alternatively, the total quantity of information (in bytes) can be determined by setting baseline = FALSE and method = 'sum'.

Author

D.E. Beaudette