Skip to content

Convert grf model to party object

Source: R/grf.R
as.party.grf.Rd

Convert a single tree from a grf (generalized random forests) model to a party object for use with partykit visualization and analysis tools.

Usage

# S3 method for class 'grf'
as.party(obj, tree = 1L, data = NULL, ...)

Arguments

obj

A grf object (e.g., regression_forest, causal_forest) from the grf package.

tree

Integer specifying which tree to convert (1-based indexing, default is 1). Must be between 1 and the number of trees in the forest.

data

Optional data.frame containing the training data. If NULL, will attempt to extract from the grf object (obj$X.orig), or create a placeholder data.frame. Providing data enables full party functionality.

...

Not currently used.

Value

A party object from the partykit package.

Details

GRF tree storage format

The grf package stores trees in a nested list structure, typically accessed via grf::get_tree(obj, tree). Each tree is represented as nested lists:

  • is_leaf: Logical, TRUE for terminal nodes

  • split_variable: 0-based index of variable to split on (internal nodes)

  • split_value: Numeric threshold for split (internal nodes)

  • left_child: Nested list for left subtree (internal nodes)

  • right_child: Nested list for right subtree (internal nodes)

  • Leaf nodes contain prediction information

Node indexing

  • Internally, grf uses 0-based variable indices

  • User-facing tree parameter uses 1-based indexing (R convention)

  • Trees use 0-based indexing internally but we access with 1-based tree number

Split encoding

  • For numeric variables: left child when feature < threshold, right child when feature >= threshold

  • partykit split created with right = TRUE (right interval closed)

Tree structure

  • grf provides nested list structure (not flattened)

  • This is the most direct representation for recursive conversion

  • Each node is a list with is_leaf flag and split info

The party object will use 1-based node IDs and variable indices as required by partykit.

Examples

if (rlang::is_installed(c("grf", "palmerpenguins"))) {
  data(penguins, package = "palmerpenguins")
  penguins <- na.omit(penguins)

  # Regression forest
  set.seed(2847)
  rf <- grf::regression_forest(
    X = penguins[, c("bill_length_mm", "bill_depth_mm",
                     "flipper_length_mm", "body_mass_g")],
    Y = penguins$bill_length_mm,
    num.trees = 3
  )

  # Convert first tree
  party_tree <- as.party(rf, tree = 1L, data = penguins)
  print(party_tree)
  plot(party_tree)

  # Can also work with other grf forest types
  set.seed(5193)
  cf <- grf::causal_forest(
    X = penguins[, c("bill_length_mm", "bill_depth_mm",
                     "flipper_length_mm", "body_mass_g")],
    Y = penguins$bill_length_mm,
    W = rbinom(nrow(penguins), 1, 0.5),
    num.trees = 3
  )
  party_tree2 <- as.party(cf, tree = 1L, data = penguins)
}
#> 
#> Model formula:
#> ~bill_length_mm + bill_depth_mm + flipper_length_mm + body_mass_g
#> 
#> Fitted party:
#> [1] root
#> |   [2] bill_length_mm <= 43.6
#> |   |   [3] bill_length_mm <= 39.2
#> |   |   |   [4] bill_length_mm <= 36.6
#> |   |   |   |   [5] bill_length_mm <= 35.5: 34.338 (n = 13, err = 10.7)
#> |   |   |   |   [6] bill_length_mm > 35.5: 36.010 (n = 20, err = 2.0)
#> |   |   |   [7] bill_length_mm > 36.6
#> |   |   |   |   [8] body_mass_g <= 3500
#> |   |   |   |   |   [9] bill_depth_mm <= 17.6: 37.814 (n = 7, err = 3.7)
#> |   |   |   |   |   [10] bill_depth_mm > 17.6: 37.809 (n = 11, err = 5.3)
#> |   |   |   |   [11] body_mass_g > 3500: 37.964 (n = 28, err = 15.9)
#> |   |   [12] bill_length_mm > 39.2
#> |   |   |   [13] bill_length_mm <= 41.6
#> |   |   |   |   [14] body_mass_g <= 3550
#> |   |   |   |   |   [15] bill_length_mm <= 39.7: 39.533 (n = 3, err = 0.0)
#> |   |   |   |   |   [16] bill_length_mm > 39.7: 40.570 (n = 10, err = 2.1)
#> |   |   |   |   [17] body_mass_g > 3550: 40.416 (n = 37, err = 20.5)
#> |   |   |   [18] bill_length_mm > 41.6
#> |   |   |   |   [19] bill_length_mm <= 42.7: 42.171 (n = 14, err = 1.3)
#> |   |   |   |   [20] bill_length_mm > 42.7: 43.129 (n = 17, err = 1.3)
#> |   [21] bill_length_mm > 43.6
#> |   |   [22] bill_depth_mm <= 19.7
#> |   |   |   [23] bill_length_mm <= 47.6
#> |   |   |   |   [24] bill_length_mm <= 45.7
#> |   |   |   |   |   [25] bill_length_mm <= 44.1: 43.800 (n = 3, err = 0.1)
#> |   |   |   |   |   [26] bill_length_mm > 44.1
#> |   |   |   |   |   |   [27] bill_length_mm <= 45.2: 44.789 (n = 9, err = 1.1)
#> |   |   |   |   |   |   [28] bill_length_mm > 45.2: 45.356 (n = 16, err = 0.3)
#> |   |   |   |   [29] bill_length_mm > 45.7
#> |   |   |   |   |   [30] bill_depth_mm <= 14.6: 46.546 (n = 13, err = 4.0)
#> |   |   |   |   |   [31] bill_depth_mm > 14.6: 46.532 (n = 31, err = 8.3)
#> |   |   |   [32] bill_length_mm > 47.6
#> |   |   |   |   [33] body_mass_g <= 4050: 50.557 (n = 23, err = 88.4)
#> |   |   |   |   [34] body_mass_g > 4050
#> |   |   |   |   |   [35] bill_length_mm <= 49.2
#> |   |   |   |   |   |   [36] bill_length_mm <= 48.7: 48.200 (n = 12, err = 1.2)
#> |   |   |   |   |   |   [37] bill_length_mm > 48.7: 48.911 (n = 9, err = 0.3)
#> |   |   |   |   |   [38] bill_length_mm > 49.2
#> |   |   |   |   |   |   [39] bill_length_mm <= 50: 49.527 (n = 11, err = 0.6)
#> |   |   |   |   |   |   [40] bill_length_mm > 50: 51.628 (n = 32, err = 133.5)
#> |   |   [41] bill_depth_mm > 19.7: 50.714 (n = 14, err = 151.3)
#> 
#> Number of inner nodes:    20
#> Number of terminal nodes: 21