update_logreg_coeffs: LR Update Function

In CCGill/LRpadding: Padded Logistic Regression

Description

Runs Logistic Regression on each response with the model and updates the coefficients, predictors, and fitted values when the log-likelihood has been improved. Updates values in place.

Usage

update_logreg_coeffs(
  model,
  responsemat,
  coeffmat,
  Xb,
  fitted,
  cols_to_update,
  padding_zeros = 0L
)

Arguments

model	- model matrix.
responsemat	matrix of response vectors in columns.
coeffmat	initial coefficient matrix.
Xb	matrix of linear predictors, one column for each response.
fitted	matrix of fitted probabilities.
cols_to_update	vector of zeros and ones (doubles) indicating which columns to update.
padding_zeros	integer number of padding zeros to add to the response and predictors to update.

Value

list of coefficient matrix, predictor matrix, fitted_values matrix, vector of old loglikelihoods (from inputs) and vector on new log likelihoods based on the returned coefficients.

CCGill/LRpadding documentation built on May 11, 2020, 12:21 a.m.