subsample_lm: Deprecated: K-fold Resampled Coefficient Estimation

In cNORM: Continuous Norming

Deprecated: K-fold Resampled Coefficient Estimation

Description

Deprecated. Averaging OLS coefficients over subsamples or folds cannot improve upon the full-sample (weighted) least squares fit, which is already the minimum-variance unbiased estimator for a fixed set of terms (Gauss-Markov); it merely adds Monte-Carlo noise and small-sample bias. The relevant source of variance is model *selection*, which is addressed by consistency screening and by BIC-weighted model averaging (bestModel(..., averaging = TRUE) or weightedAverageModel). This function now simply returns the full-sample least squares fit.

Usage

subsample_lm(text, data, weights, k = 10)

Arguments

text	A character string or formula specifying the model to be fitted
data	A data frame containing the variables in the model
weights	Optional numeric vector of weights
k	Ignored (kept for backwards compatibility)

Value

An object of class 'lm' fitted on the complete sample.

See Also

weightedAverageModel

cNORM documentation built on July 13, 2026, 5:08 p.m.