stepwise.backward {DGM}

Stepise backward non-exhaustive greedy search, calculates the optimum value of the discount factor.

Description

Stepise backward non-exhaustive greedy search, calculates the optimum value of the discount factor.

Usage

stepwise.backward(
  Data,
  node,
  nbf = 15,
  delta = seq(0.5, 1, 0.01),
  max.break = TRUE,
  priors = priors.spec()
)

Arguments

Data

Dataset with dimension number of time points T x number of nodes Nn.

node

The node to find parents for.

nbf

The Log Predictive Likelihood will sum from (and including) this time point.

delta

A vector of values for the discount factor.

max.break

If TRUE, the code will break if adding / removing parents does not improve the LPL. If FALSE, the code will continue to the zero parent / all parent model. Default is TRUE.

priors

List with prior hyperparameters.

Value

model.store The parents, LPL and chosen discount factor for the subset of models scored using this method.