hdcd_control: Create an object of class hdcd_control to supply parameters...
In mlondschien/hdcd: High - Dimensional Change Point Detection

Create an object of class hdcd_control to supply parameters to hdcd

hdcd_control(glasso_cv_inner = TRUE, glasso_cv_inner_lambda = NULL,
  glasso_cv_inner_randomize_folds = FALSE, glasso_cv_inner_n_folds = 5,
  glasso_cv_inner_lambda_stepsize = 0.1^0.1,
  glasso_NA_method = "loh_wainwright_bias_correction",
  glasso_standardize = TRUE, glasso_penalize_diagonal = FALSE,
  classifier_oob = TRUE, section_search_min_points = 5,
  section_search_stepsize = 0.5, random_forest_n_tree = 600,
  random_forest_mtry = NULL, random_forest_sample_fraction = 1,
  kNN_k = function(x) ceiling(sqrt(x)), wbs_n_segments = 100,
  sbs_alpha = 1/sqrt(2), permutation_test = FALSE,
  permutation_test_pvalue = 0.05, permutation_test_n = 400)

`glasso_cv_inner`	Should change points be selected according to inner cross validation in the glassocd procedure.
`glasso_cv_inner_lambda`	Grid of values for lambda to be analyzed during inner cross validation. If NULL, values will be searched for iteratively
`glasso_cv_inner_randomize_folds`	Should inner folds be selected in a randomized fashion (instead of equispaced)
`glasso_cv_inner_n_folds`	Number of inner folds
`glasso_NA_method`	Method to be used for covariance estimation in the presence of missing values. One of `loh_wainwright_bias_correction`, `pairwise` or `average`. See accompanying paper for more details.
`glasso_standardize`	Should values be be standardized before they are fed into the glasso?
`glasso_penalize_diagonal`	Should values on the diagonal of the precision matrix be penalized in the glasso?
`classifier_oob`	Does the (custom) classifier return oob predictions?
`section_search_stepsize`	Stepsize parameter for the section_search optimizer
`random_forest_n_tree`	n_tree parameter for Random Forest algorithm ranger
`random_forest_mtry`	mtry parameter for Random Forest algorithm ranger
`random_forest_sample_fraction`	sample_fraction parameter for Random Forest algorithm ranger
`kNN_k`	Number of nearest neighbors used for classification. Either a positive integer or a function that reuturns a positive integer given the number of total observations.
`wbs_n_segments`	Number of segments to be drawn for the WBS procedure
`sbs_alpha`	Decay parameter for the SBS procedure
`permutation_test`	should a permutation test be done (approximated) for model selection (only relevant for kNNcd and RFcd)
`permutation_test_pvalue`	pvalue threshold used in permutation test
`permutation_test_n`	number of permutations (approximations) in permutation test