checkCI: Check conditional independence

In rje42/contingency: Discrete Multivariate Probability Distributions

Check conditional independence

Description

Gives a numerical check that a (conditional) independence holds in a probability distribution.

Usage

checkCI(x, A, B, C = integer(0), eps = .Machine$double.eps, ...)

## S3 method for class 'array'
checkCI(x, A, B, C = integer(0), eps = .Machine$double.eps, ...)

## S3 method for class 'tables'
checkCI(x, A, B, C = integer(0), eps = .Machine$double.eps, ...)

Arguments

x	an array or object of class tables
A, B	the sets of variables whose independence is to be tested
C	conditioning set (possibly empty)
eps	tolerance parameter
...	other arguments to methods

Details

just tests to an appropriate numerical precision that a conditional independence holds: this is not a statistical test for conditional independence. If A and B overlap with C then these vertices are ignored. If A and B intersect with one another (but not C) then the solution is always false.

Value

A logical, or a vector of logicals of the same length as the number of tables provided, indicating whether the conditional independence seems to hold numerically.

Methods (by class)

• checkCI(array): method for array object

• checkCI(tables): method for tables object

rje42/contingency documentation built on Sept. 9, 2024, 6:32 p.m.