random_sevt {stagedtrees}

Generate a random (fitted) sevt

Description

Generate a random sevt from a DAG or a tree. Probabilities are also randomly generated.

Usage

random_sevt(x, q = 0.5, rfun = rexp)

## S3 method for class 'list'
random_sevt(x, q = 0.5, rfun = rexp)

## S3 method for class 'parentslist'
random_sevt(x, q = 0.5, rfun = rexp)

## S3 method for class 'sevt'
random_sevt(x, q = 0.5, rfun = rexp)

Arguments

x

a sevt object, a parentslist object or a list.

q

probability of joining stages.

rfun

a function which is used to generate random conditional probabilities associated to each stage.

Details

The generated staged tree is obtained by randomly joining stages with probability q.

For random_sevt.list, x should be a list representing an event tree, same format as lists provided to sevt.list. The random generated sevt will be obtained by randomly joining stages starting from a full staged event tree.

For random_sevt.parentslist, x should be a parentslist object representing a DAG, this could be obtained with as_parentslist or with random_parentslist. The random generated sevt will be obtained by randomly joining stages starting from a the staged tree equivalent to the DAG.

For random_sevt.sevt, x should be a sevt. The random generated sevt will be obtained by randomly joining stages starting from the provided sevt object.

     Stages (conditional) probabilities are sampled from
     the corresponding probability simplex by generating
     a vector with the user-defined function \code{rfun} and
     normalizing it to sum up to one.
     Absolute value is applied to assure non-negativity.
     The default \code{rfun = rexp} induces a uniform sampling
     from the probability simplex.

Value

A randomly generated fitted sevt object.

Examples

model_gt <- random_sevt(list(
  X = c("a", "b"), Y = c("c", "d", "e"),
  Z = c("1", "2", "3"), W = c("yes", "no")
))

## sample data from model_gt and estimate a staged tree
data <- sample_from(model_gt, 100)
model_est <- stages_bhc(full(data))

## compare true and estimated model
hamming_stages(model_gt, model_est)
compare_stages(model_gt, model_est, method = "hamming", plot = TRUE)