biC.ADMM.speed: biC-ADMM: Biclustering Algorithm for Model with Compositional...

In sakuramomo1005/biADMM: Convex bicluster alogrithm based on ADMM algorithm

Description

Same algorithm as biC.ADMM. Call python code to speed up the running time.

Usage

biC.ADMM.speed(
  X,
  nu1,
  nu2,
  nu3,
  gamma_1,
  gamma_2,
  m,
  phi,
  prox = "l2",
  niters = 10,
  tol = 0.1,
  weight.scale = 1,
  output = 1
)

Arguments

X	The data matrix to be clustered. The rows are the samples, and the columns are the features.
nu1	A regularization parameter for row shrinkage
nu2	A regularization parameter for column shrinkage
nu3	A regularization parameter for compositional data constrain
gamma_1	A regularization parameter for row shrinkage
gamma_2	A regularization parameter for column shrinkage
m	m-nearest-neighbors in the weight function
phi	The parameter phi in the weight function
prox	The proximal maps. Could calculate L1 norm, L2 norm, or L-infinity, use "l1", "l2", or "l-inf", respectively.
niters	Iteraion times
tol	Stopping criterion
weight.scale	If weight.scale = 1, the code will make the input data have compositional structure.
output	When output = 1, print the results at each iteration. No print when output equals other value.

Value

A list of results, containing matrix of A, v, z, lambda1, lambda2, and lambda3

Examples

# generate dataset
set.seed(123)
X = data_gen(n = 100, p = 80)
# set parameters
nu1 = nu2 = nu3 = gamma_1 = gamma_2 = 0.1
m = 5
phi = 0.5
# biADMM algorithm
res4 = biC.ADMM.speed(X, nu1, nu2, nu3, gamma_1, gamma_2,
 m, phi, niter = 10, tol = 0.0001, weight.scale = 1, output = 0)
dim(res4$A)

sakuramomo1005/biADMM documentation built on March 25, 2021, 1:38 p.m.