LinearModelL1: LinearModelL1

In ChaddFrasier/planetLearn: Planetary Learning Package

Description

minimizes the following cost function: 1/n ∑i=1^n L[w^T x_i + b, y_i] + penalty * ||w||_1

Usage

LinearModelL1(
  X.scaled.mat,
  y.vec,
  penalty,
  opt.thresh,
  initial.weight.vec,
  step.size
)

Arguments

X.scaled.mat	the scaled training matrix [ n_observations : n_features ]
y.vec	the label vector for the data matrix [ n_observations : 1 ]
penalty	a positive scaler penalty term
opt.thresh	a positive scaler
initial.weight.vec	a starting weight vector, can be random numbers or 0s [ n_features : 1 ]
step.size	the incrimental step to descent with

Value

an optimal weight vector for predictings [n_features + 1 : 1] where the first entry is the intercept term

Examples

# load the default dataset
getwd()

ChaddFrasier/planetLearn documentation built on July 5, 2020, 2:32 a.m.