stagedtrees (version 2.3.0)

prob: Probabilities for a staged event tree

Description

Compute (marginal and/or conditional) probabilities of elementary events with respect to the probability encoded in a staged event tree.

Usage

prob(object, x, conditional_on = NULL, log = FALSE, na0 = TRUE)

Value

the probabilities to observe each observation in x, possibly conditional on the event(s) in conditional_on.

Arguments

object

an object of class sevt with probabilities.

x

the vector or data.frame of observations.

conditional_on

named vector, the conditioning event.

log

logical, if TRUE log-probabilities are returned.

na0

logical, if NA should be converted to 0.

Details

Computes probabilities related to a vector or a data.frame of observations.

Optionally, conditional probabilities can be obtained by specifying the conditioning event in conditional_on. This can be done either with a single named vector or with a data.frame object with the same number of rows of x. In the former, the same conditioning is used for all the computed probabilities (if x has multiple rows); while with the latter different conditioning events (but on the same variables) can be specified for each row of x.

Examples

data(Titanic)
model <- full(Titanic, lambda = 1)
samples <- expand.grid(model$tree[c(1, 4)])
pr <- prob(model, samples)
## probabilities sum up to one
sum(pr)
## print observations with probabilities
print(cbind(samples, probability = pr))

## compute one probability
prob(model, c(Class = "1st", Survived = "Yes"))

## compute conditional probability
prob(model, c(Survived = "Yes"), conditional_on = c(Class = "1st"))

## compute conditional probabilities with different conditioning set
prob(model, data.frame(Age = rep("Adult", 8)),
  conditional_on = expand.grid(model$tree[2:1])
)
## the above should be the same as
summary(model)$stages.info$Age