epsgo: Efficient Parameter Selection via Global Optimization

In zhizuio/mixlasso: Mix lasso models

Efficient Parameter Selection via Global Optimization

Description

Finds an optimal solution for the Q.func function.

Usage

epsgo(
  Q.func,
  bounds,
  bound.scale = NA,
  x,
  y,
  z = z,
  family = "gaussian",
  lambda = NULL,
  alpha = 1,
  p = NULL,
  intercept = TRUE,
  foldid = NULL,
  nfolds = 10,
  cv.measure = NULL,
  type.min = "lambda.min",
  tree.parm = tree.parm,
  num.nonpen = 0,
  strata.surv = NULL,
  threshold = 0,
  mu = 0.01,
  NoVar = 50,
  standardize.response = FALSE,
  round.n = 5,
  parms.coding = "none",
  fminlower = 0,
  flag.find.one.min = FALSE,
  show = "none",
  N = NULL,
  maxevals = 500,
  constantMean = 0,
  epsilon = 1e-04,
  Dir.ep = 1e-04,
  Dir.tol = 0.01,
  EI.eps = 0.01,
  min.iter = 10,
  pdf.name = NULL,
  pdf.width = 12,
  pdf.height = 12,
  my.mfrow = c(1, 1),
  parallel = FALSE,
  modelList = NULL,
  verbose = TRUE,
  seed = 123,
  search.path = FALSE,
  tol = tol,
  y.mis = y.mis,
  x.mis = x.mis,
  t.idx = t.idx,
  t.glasso = FALSE,
  maxiter = 10000,
  cov.proxy = "FL",
  predict.re = FALSE,
  ...
)

Arguments

Q.func	name of the function to be minimized.
bounds	bounds for the interval-searching parameters
x, y	input matrix.
family	response type.
lambda	optional user-supplied lambda sequence; default is NULL, and espsgo chooses its own sequence.
p	the numbers of predictors from different data sources.
intercept	should intercept(s) be fitted (default=TRUE) or set to zero (FALSE).
foldid	an vector of values for the cross-validation.
num.nonpen	number of predictors forced to be estimated (i.e., nonpenalization).
strata.surv	stratification variable for the Cox survival model.
threshold	threshold for estimated coefficients of the tree-lasso models.
standardize.response	standardization for the response variables. Default: TRUE.
round.n	number of digits after comma, default is 5.
parms.coding	parmeters coding: none or log2, default: none.
fminlower	minimal value for the function Q.func, default is 0.
flag.find.one.min	do you want to find one min value and stop? Default: FALSE.
show	show plots of DIRECT algorithm: none, final iteration, all iterations. Default: none.
N	define the number of start points depending on the dimensionality of the parameter space.
maxevals	the maximum number of DIRECT function evaluations, default: 500.
EI.eps	the convergence threshold for the expected improvement between fmin and the updated point
min.iter	the minimus iterations after the initial N iterations.
parallel	If TRUE, use parallel foreach to fit each fold except parallelizing each lambda for the tree-lasso methods. If c(TRUE,TRUE), use parallel foreach to fit each fold and each lambda.
modelList	detailed information of the search process
verbose	print the middle search information, default is TRUE.
seed	random seed
search.path	save the visited points, default is FALSE.
espilon	the convergence shreshold for the function Q.func, default is 0.01.

Details

mixlasso

References

Frohlich, H. & Zell, A. (2005). Efficient Parameter Selection for Support Vector Machines in Clas- sification and Regression via Model-Based Global Optimization. Proceedings of the International Joint Conference of Neural Networks, pp 1431-1438.

Sill, M., Hielscher, T., Becker, N. & Zucknick, M. (2014).c060: Extended Inference with Lasso and elastic net Regularized Cox and Generalized Linear methods. Journal of Statistical Software, 62(5):1-22.

zhizuio/mixlasso documentation built on March 21, 2022, 1:07 a.m.