Guess the best normalization type from the stable interior of a distribution

Applies heuristics based on the support range, boundary mass, and zero inflation to select among "asinh", "logit", "rank_normal", and "hurdle".

Usage

guess_normalize_type(
  x,
  range,
  boundary_mass_cutoff = 0.1,
  zero_tol = 1e-08,
  one_tol = 1e-08
)

Arguments

x

Numeric vector of stable (non-sentinel, non-NA) winsorized values.

range

Character range code used to clip the vector to its theoretical support before assessment.

boundary_mass_cutoff

Proportion of values at or near zero (or one) that triggers the "hurdle" or "rank_normal" path. Default 0.10.

zero_tol

Tolerance for identifying values near zero. Default 1e-8.

one_tol

Tolerance for identifying values near one. Default 1e-8.

Value

A named list with elements type (character) and reason (character explanation).