linear_gd_optim2: Linear Gradient Descend (2)

In FedericoCortese/R4DScm:

Description

This function is the slower version of linear_gd_optim given that it makes use of the grad function contained in the package "numDeriv".

Usage

linear_gd_optim2(b_pre, X, y, tol = 0.001, maxit = 1000, stepsize = 0.001)

Arguments

b_pre	vector of initial parameters
X	Covariates Matrix: each column contains observations for each covariate.
y	Response variable observations
tol	Tolerance level for the optimization process, the default is 0.001.
maxit	Maximum iterations number. Default is 1000.
stepsize	The value for the stepsize in the equation of the gradient descend. Default is 0.001.

Value

A list containing the fitted values for the beta vector and the number of iterations performed.

Examples

set.seed(8675309)
n = 1000
x1 = rnorm(n)
x2 = rnorm(n)
y = 1 + .5*x1 + .2*x2 + rnorm(n)
X=cbind(x1,x2)
b_pre=c(0,0,0)
linear_gd_optim(b_pre,X,y)

FedericoCortese/R4DScm documentation built on July 12, 2021, 5:30 a.m.