Substitution-metric distance between distributions

tda_subm measures the substitution-metric distance between a reference distribution and one or more others, TDA's subm.

Usage

tda_subm(..., options = list(), dir = tempfile("tda"))

Arguments

...

the distribution variables: the reference first, then one or more to compare against it.

options

a named list of further TDA options, passed through.

dir

working directory.

Value

An object carrying distances, a named numeric vector.

Details

Its columns are not raw data: every value must lie in [0, 1], one row per support point, each column a probability distribution over the same support – TDA rejects anything else with “no distribution in variable”. The first column is the reference; the rest are compared against it, giving one distance each.

TDA prints the distances rather than writing them to the file df= names (that file holds the substitution detail, not the summary), so tda_subm parses them from the run's output into distances, a named numeric vector – table stays NULL unless options = list(df = ...) is given, in which case it holds that detail instead.

See also

Other descriptive statistics: tda_brr(), tda_cov(), tda_dstat(), tda_freq(), tda_independence(), tda_ineq(), tda_loglin(), tda_quant(), tda_rcorr(), tda_segr()

Examples

# six points of support (say, a 6-point agreement scale): Y1 is close
# to the reference X, Y2 is a very different shape
d <- data.frame(X  = c(0.05, 0.15, 0.30, 0.30, 0.15, 0.05),
                 Y1 = c(0.03, 0.12, 0.32, 0.33, 0.15, 0.05),
                 Y2 = c(0.30, 0.30, 0.20, 0.10, 0.07, 0.03))
tda_subm(d)$distances
#>   Y1   Y2 
#> 0.10 1.07