How far two integer-valued variables are from independent: for each
subset of y's values, the largest gap between the distribution
of x within that subset and the overall distribution of
x – TDA's indep. A delta of 0 means x is
distributed identically inside and outside the subset.
Usage
tda_independence(
x,
y,
data = NULL,
partition = NULL,
dir = tempfile("tda"),
...
)Arguments
- x, y
integer-valued vectors of equal length, or column names in
data.- data
optional data frame supplying
xandy.- partition
optional increasing integers: the upper ends of consecutive
ysubsets.- dir
working directory.
- ...
passed to
tda_run.
Value
A data frame with one row per subset: subset (its
number), values (the y values it contains, as a
comma-separated string), and delta.
Details
By default every distinct value of y is its subset. With
partition, y's range is cut into consecutive subsets:
each number is the last y value of one subset (TDA's
y=). Values must be integers; TDA works on the integer grid
from the smallest to the largest observed value.
See also
Other descriptive statistics:
tda_brr(),
tda_cov(),
tda_dstat(),
tda_freq(),
tda_ineq(),
tda_loglin(),
tda_quant(),
tda_rcorr(),
tda_segr(),
tda_subm()
Examples
d <- data.frame(x = c(1, 2, 3, 4, 1, 2, 3, 4),
y = c(1, 1, 1, 1, 2, 2, 2, 2))
tda_independence("x", "y", data = d)
#> subset values delta
#> 1 1 1 0
#> 2 2 2 0