Find the globally best greedy concentration-index split — best_global_ci

Search across all supplied variables and their admissible split points, then return the variable-split pair with the largest concentration-index gain.

Usage

best_global_ci_split(x, y, wt, ctrl, ci_fun, vars = seq_along(x))

Arguments

x: A data frame or list of predictors.
y: A two-column numeric response matrix.
wt: Case weights.
ctrl: A control list created by ci_tree_control() or compatible.
ci_fun: Concentration-index scoring function.
vars: Optional integer variable ids to search. Defaults to all columns of x.

Value

A split candidate list, or NULL when no variable has an admissible gain above control$min_gain and parent-node relative gain above control$min_relative_gain.