evidence_object: Evidence objects

In hojsgaard/gRain: Graphical Independence Networks

Evidence objects

Description

Functions for defining and manipulating evidence.

Usage

new_evi(evi_list = NULL, levels)

is.null_evi(object)

## S3 method for class 'grain_evidence'
as.data.frame(x, row.names = NULL, optional = FALSE, ...)

setdiff_evi(ev1, ev2)

union_evi(ev1, ev2)

Arguments

evi_list	A named list with evidence; see 'examples' below.
levels	A named list with the levels of all variables.
object	Some R object.
x	An evidence object.
row.names	Not used.
optional	Not used.
...	Not used.
ev1, ev2	Evidence.

Details

Evidence is specified as a list. Internally, evidence is represented as a grain evidence object which is a list with 4 elements.

Author(s)

Søren Højsgaard, [e-mail]

Examples

## Define the universe
yn <- c("yes", "no")
uni <- list(asia = yn, tub = yn, smoke = yn, lung = yn,
            bronc = yn, either = yn, xray = yn, dysp = yn)

e1 <- list(dysp="no", xray="no")
eo1 <- new_evi(e1, levels=uni)
eo1  |> as.data.frame()

e2 <- list(dysp="no", xray=c(0, 1))
eo2 <- new_evi(e2, levels=uni)
eo2 |> as.data.frame()

# Above e1 and e2 specifies the same evidence but information about
# whether the state has been set definite or as a weight is
# maintained.

e3 <- list(dysp="yes", asia="yes")
eo3 <- new_evi(e3, uni)
eo3 |> as.data.frame()

# If evidence 'e1' is already set in the network and new evidence
# 'e3' emerges, then evidence in the network must be updated. But
# there is a conflict in that dysp="yes" in 'e1' and
# dysp="no" in 'e3'. The (arbitrary) convention is that
# existing evidence overrides new evidence so that the only new
# evidence in 'e3' is really asia="yes".

# To subtract existing evidence from new evidence we can do:
setdiff_evi(eo3, eo1) |> as.data.frame()

# Likewise the 'union' is
union_evi(eo3, eo1) |> as.data.frame()

hojsgaard/gRain documentation built on Jan. 10, 2024, 9:27 p.m.