GFCI Algorithm for Causal Discovery

Run the GFCI (Greedy Fast Causal Inference) algorithm for causal discovery using one of several engines. This combines the FGES and FCI algorithms.

Usage

gfci(engine = "tetrad", score, test, alpha = 0.05, ...)

Arguments

engine

Character; which engine to use. Must be one of:

"tetrad"

Tetrad Java library.

score

Character; name of the scoring function to use.

test

Character; name of the conditional‐independence test.

alpha

Numeric; significance level for the CI tests.

...

Additional arguments passed to the chosen engine (e.g. score and algorithm parameters).

Details

For specific details on the supported scores, and parameters for each engine, see:

• TetradSearch for Tetrad.

Recommendation

While it is possible to call the function returned directly with a data frame, we recommend using disco(). This provides a consistent interface and handles knowledge integration.

Value

A function that takes a single argument data (a data frame). When called, this function returns a list containing:

• knowledge A Knowledge object with the background knowledge used in the causal discovery algorithm. See knowledge() for how to construct it.

• caugi A caugi::caugi object representing the learned causal graph. This graph is a PAG (Partial Ancestral Graph), but since PAGs are not yet natively supported in caugi, it is currently stored with class UNKNOWN.

See also

Other causal discovery algorithms: boss(), boss_fci(), fci(), ges(), grasp(), grasp_fci(), gs(), iamb-family, pc(), sp_fci(), tfci(), tges(), tpc()

Examples

data(num_data)

# Requires Tetrad to be installed
if (verify_tetrad()$installed && verify_tetrad()$java_ok) {
  # Recommended path using disco()
  gfci_tetrad <- gfci(
    engine = "tetrad",
    score = "sem_bic",
    test = "fisher_z"
  )
  disco(tpc_example, gfci_tetrad)

  # or using gfci_tetrad directly
  gfci_tetrad(tpc_example)
}
#> 
#> ── caugi graph ─────────────────────────────────────────────────────────────────
#> Graph class: UNKNOWN
#> 
#> ── Edges ──
#> 
#>   from      edge  to       
#>   <chr>     <chr> <chr>    
#> 1 child_x2  o-o   child_x1 
#> 2 child_x2  o->   oldage_x5
#> 3 child_x2  o-o   youth_x4 
#> 4 oldage_x5 -->   oldage_x6
#> 5 youth_x3  o->   oldage_x5
#> 6 youth_x4  -->   oldage_x6
#> ── Nodes ──
#> 
#>   name     
#>   <chr>    
#> 1 child_x2 
#> 2 child_x1 
#> 3 youth_x4 
#> 4 youth_x3 
#> 5 oldage_x6
#> 6 oldage_x5
#> ── Knowledge object ────────────────────────────────────────────────────────────

#### With tier knowledge ####
if (verify_tetrad()$installed && verify_tetrad()$java_ok) {
  kn <- knowledge(
    tpc_example,
    tier(
      child ~ tidyselect::starts_with("child"),
      youth ~ tidyselect::starts_with("youth"),
      oldage ~ tidyselect::starts_with("oldage")
    )
  )

  # Recommended path using disco()
  gfci_tetrad <- gfci(
    engine = "tetrad",
    score = "sem_bic",
    test = "fisher_z"
  )
  disco(tpc_example, gfci_tetrad, knowledge = kn)

  # or using gfci_tetrad directly
  gfci_tetrad <- gfci_tetrad |> set_knowledge(kn)
  gfci_tetrad(tpc_example)
}
#> 
#> ── caugi graph ─────────────────────────────────────────────────────────────────
#> Graph class: UNKNOWN
#> 
#> ── Edges ──
#> 
#>   from      edge  to       
#>   <chr>     <chr> <chr>    
#> 1 child_x2  o-o   child_x1 
#> 2 child_x2  o->   oldage_x5
#> 3 child_x2  o-o   youth_x4 
#> 4 oldage_x5 -->   oldage_x6
#> 5 youth_x3  o->   oldage_x5
#> 6 youth_x4  -->   oldage_x6
#> ── Nodes ──
#> 
#>   name     
#>   <chr>    
#> 1 child_x2 
#> 2 child_x1 
#> 3 youth_x4 
#> 4 youth_x3 
#> 5 oldage_x6
#> 6 oldage_x5
#> ── Knowledge object ────────────────────────────────────────────────────────────

# With all algorithm arguments specified
if (verify_tetrad()$installed && verify_tetrad()$java_ok) {
  gfci_tetrad <- gfci(
    engine = "tetrad",
    score = "poisson_prior",
    test = "rank_independence",
    depth = 3,
    max_degree = 2,
    max_disc_path_length = 5,
    use_heuristic = FALSE,
    complete_rule_set_used = FALSE,
    guarantee_pag = TRUE,
    start_complete = TRUE,
    num_threads = 2,
    verbose = TRUE
  )
  disco(num_data, gfci_tetrad)
}
#> 
#> ── caugi graph ─────────────────────────────────────────────────────────────────
#> Graph class: UNKNOWN
#> 
#> ── Nodes ──
#> 
#>   name 
#>   <chr>
#> 1 X1   
#> 2 X2   
#> 3 X3   
#> 4 Z    
#> 5 Y    
#> ── Knowledge object ────────────────────────────────────────────────────────────