gradient_descent: Gradient descent for the 'LocationScaleRegression' model...

In hriebl/asp20model: An R6 Class for Location-Scale Regression Models

Description

This function optimizes the log-likelihood of the given location-scale regression model by gradient descent. It has a side effect on the model object.

Usage

gradient_descent(
  model,
  stepsize = 0.001,
  maxit = 1000,
  abstol = 0.001,
  verbose = FALSE
)

Arguments

model	A LocationScaleRegression object.
stepsize	The scaling factor for the gradient.
maxit	The maximum number of iterations.
abstol	The absolute convergence tolerance. The algorithm stops if the absolute value of the gradient drops below this value.
verbose	Whether to print the progress of the algorithm.

Value

The updated model, invisibly.

Examples

y <- rnorm(30)
model <- LocationScaleRegression$new(y ~ 1)
gradient_descent(model)

hriebl/asp20model documentation built on April 2, 2020, 12:05 a.m.