Build terminal-node summary statistics for a fitted tree

Computes one-row summaries for each terminal node of a fitted tree by first assigning observations to nodes and then applying user-supplied summary functions within each terminal group.

Usage

tree_build_terminal_stats(fit, data, stat_funs, node_col = "node_id")

Arguments

fit

A fitted tree object inheriting from party.

data

A data frame used to assign observations to terminal nodes.

stat_funs

A named list of functions, each returning a single summary value for a node-specific data frame.

node_col

Name of the output column storing terminal node ids.

Value

A data frame with one row per terminal node and one column per summary statistic.

Examples

data(kenya, package = "ineqTrees")
kenya_plot_vars <- c("wealth", "deadu5_num", "rural", "ed", "reg", "unskilled")
kenya_plot_data <- kenya[
  stats::complete.cases(kenya[, kenya_plot_vars]),
  kenya_plot_vars
]
set.seed(20260512)
kenya_plot_data <- kenya_plot_data[
  sample.int(nrow(kenya_plot_data), 800L),
  ,
  drop = FALSE
]
kenya_plot_fit <- ci_tree(
  cbind(wealth, deadu5_num) ~ rural + ed + reg + unskilled,
  data = kenya_plot_data,
  rank_name = "wealth",
  outcome_name = "deadu5_num",
  control = ci_tree_control(
    minsplit = 100,
    minbucket = 50,
    minprob = 0.05,
    maxdepth = 3
  )
)
kenya_stat_funs <- list(
  n = nrow,
  mortality = function(df) mean(df$deadu5_num),
  mean_wealth = function(df) mean(df$wealth)
)

tree_build_terminal_stats(
  kenya_plot_fit,
  kenya_plot_data,
  stat_funs = kenya_stat_funs
)
#>   node_id   n  mortality mean_wealth
#> 1       4 109 0.00000000  0.10045582
#> 2       5 132 0.02272727  0.31801860
#> 3       6 137 0.02919708 -0.23154495
#> 4       9 209 0.04784689  0.04196376
#> 5      10 127 0.15748031 -0.38363010
#> 6      11  86 0.24418605 -0.90504013