permfit: PermFIT: A permutation-based feature importance test.

In SkadiEye/deepTL: Deep Treatment Learning

PermFIT: A permutation-based feature importance test.

Description

PermFIT: A permutation-based feature importance test.

Usage

permfit(
  train,
  validate = NULL,
  k_fold = 5,
  n_perm = 100,
  pathway_list = list(),
  method = c("ensemble_dnnet", "random_forest", "lasso", "linear", "svm", "dnnet",
    "xgboost")[1],
  shuffle = NULL,
  ...
)

Arguments

train	An dnnetInput or dnnetSurvInput object.
validate	A validation dataset is required when k_fold = 0.
k_fold	K-fold cross-fitting. If not, set k_fold to zero.
n_perm	Number of permutations repeated.
pathway_list	A list of pathways to be jointly tested.
method	Models, including ensemble_dnnet for ensemble deep neural networks, random_forest for random forests or random survival forests, lasso for linear/logistic/cox lasso, \ linear for linear/logistic/coxph regression, svm for svms with Gaussian kernels, and dnnet with single deep neural network.
shuffle	If shuffle is null, the data will be shuffled for cross-fitting; if random shuffle is not desired, please provide a bector of numbers for cross-fitting indices
...	Additional parameters passed to each method.

Value

Returns a PrmFIT object.

SkadiEye/deepTL documentation built on Nov. 17, 2022, 1:41 p.m.