linear_gd_optim: linear_gd_optim

In lucaaiello/myOpt: My Optimization Package

Description

This function provides a personalized efficient way for optimizing functions. It is used to fit linear models, applying the gradient descent method.

Usage

linear_gd_optim(
  par,
  X,
  Y,
  tolerance = 1e-06,
  maxit = 10000,
  stepsize = 0.001,
  verbose = T
)

Arguments

par	Initial values.
X	Data matrix predictors (first column has to contain values 1 to consider the intercept).
Y	Response vector.
tolerance	Value for the stopping criterion. The default value is set to 1e-6.
maxit	Maximum iterations allowed. The default value is set to 10000.
stepsize	Length of the stepsize parameter. The default value is set to 1e-3.
verbose	If set TRUE the function produce messages during the computation.

Value

Returns the vector containing the estimated parameters.

Examples

library(myOpt)
## basic example code

set.seed(8675309)

# data simulation for example purposes

n = 1000

x1 = rnorm(n)
x2 = rnorm(n)
X <- cbind(rep(1,n),x1,x2)

Y = 1 + 0.5*x1 + 0.2*x2 + rnorm(n)

# random initial values for the parameters of the linear model

par <- rnorm(dim(X)[2])


est_par <- linear_gd_optim(par, X, Y)

lucaaiello/myOpt documentation built on Dec. 21, 2021, 11:51 a.m.