gscca.simu: Simulation for gscca

In DongyueXie/scca: Sparse Canonical Correlation Analysis - Intergrated Methods

Description

This function generate tree from data each time instead of using same tree for all simulations

Usage

gscca.simu(niter, u, v, n = 50, data.type = "normal", sigma_z = 1,
  sigma_x = NULL, sigma_y = NULL, theta = 0.01, lambda_z = 10,
  n_x = 1000, n_y = 1000, cv.method = "bic", gamma.u = NULL,
  gamma.v = NULL, edgex = NULL, edgey = NULL, Sx = NULL, Sy = NULL,
  p_c = 1, dis = F, p_d = 1, h = 1, wcorr = T, thresh = 0.5,
  lambda.u = NULL, lambda.v = NULL, maxsteps = 20, plain = T,
  verbose = F, K = 5, ccor = F, seed = 123)

Arguments

gamma.u, gamma.v:	ratio of l1 peanlty and fused penalty
Sx, Sy:	structure matrix for Sx,Sy.
lambda.u, lambda.v:	fused penalty. lambda*fused penalty+gamma*lambda*l1 penalty.
maxsteps:	the number of lambdas to try, default to be 20. Largest 20 values from solution path
ccor:	whether re-estiamte the coeeficient when doing BIC

DongyueXie/scca documentation built on May 29, 2019, 2:37 p.m.