approximate trend filtering via MM algorithm

Share:

Description

Uses majorization-minimization technique to approximate trend filtering fit.

Usage

1
2
tf(y, x = NULL, k = 2, l, D = NULL, eps = 1e-08, tau = 1e-05,
  max_iter = 500)

Arguments

y

observed data

x

inputs corresponding to observations

k

order of fit

l

vector of penalty parameters lambda

D

matrix Delta of order k+1

eps

error adjustment to majorization function

tau

convergence threshold

max_iter

maximum number of iterations allowed

Author(s)

Edward A. Roualdes