make_partitions: Create Data Partitions Using Clustering

In lgspline: Lagrangian Multiplier Smoothing Splines for Smooth Function Estimation

Create Data Partitions Using Clustering

Description

Partitions data support into clusters using Voronoi-like diagrams. Standardization for clustering is handled internally; centers, knots, and the returned assign_partition function are all on the **raw (natural) scale** of the original predictors.

Usage

make_partitions(
  data,
  cluster_args,
  cluster_on_indicators,
  K,
  parallel,
  cl,
  do_not_cluster_on_these,
  neighbor_tolerance,
  standardize = TRUE,
  standardize_mode = "auto",
  dummy_adder = 0,
  dummy_dividor = 0
)

Arguments

data	Numeric matrix of predictor variables (raw scale)
cluster_args	Parameters for clustering
cluster_on_indicators	Logical to include binary predictors
K	Number of partitions minus 1 (K)
parallel	Logical to enable parallel processing
cl	Cluster object for parallel computation
do_not_cluster_on_these	Columns to exclude from clustering
neighbor_tolerance	Scaling factor for neighbor detection
standardize	Logical; whether to standardize data internally before clustering. Should equal standardize_predictors_for_knots from lgspline.fit. Default TRUE.
standardize_mode	Character; "auto" (default) selects "minmax" for a single effective clustering dimension and "normal" for multiple dimensions. Can be forced to either value.
dummy_adder	Small constant added to numerator during standardization to avoid exact-zero values (matches lgspline.fit's dummy_adder). Default 0.
dummy_dividor	Small constant added to denominator during standardization to avoid division by zero (matches lgspline.fit's dummy_dividor). Default 0.

Value

A list containing:

centers

Cluster center coordinates on the raw scale.

knots

Knot points between centers on the raw scale.

assign_partition

Function that accepts raw-scale new data and returns integer-like partition assignments (0.5, 1.5, ...).

neighbors

List of neighboring partition indices.

standardize_transf

The forward standardization function used internally (for diagnostic use only).

standardize_inv_transf

The inverse standardization function (for diagnostic use only).

centers_std

Cluster centers on the standardized scale (for diagnostic use only).

lgspline documentation built on May 8, 2026, 5:07 p.m.