best_predictor_binomial: Using bootstrapping and LASSO algorithm to choose best...

In DongqiangZeng0808/Blasso: Using LASSO and Bootstrapping algorithm to evaluate predictive value of model

Using bootstrapping and LASSO algorithm to choose best predictive features

Description

Using bootstrapping and LASSO algorithm to choose best predictive features

Usage

best_predictor_binomial(
  target_data,
  response = "response",
  features,
  target_data_id = "ID",
  features_id = "ID",
  show_progress = TRUE,
  permutation = 1000,
  propotion = 0.8,
  nfolds = 10,
  plot_vars = 20,
  color = "#999999",
  palette = "Blues",
  discrete_x = 20
)

Arguments

target_data	Data frame contains patient identifier, and target_data
response	binary variables
features	Data frame contains patient identifier and variables after feature engineering
target_data_id	identifier of phenotype data
features_id	identifier of feature matrix
show_progress	show progress bar
permutation	times of iteration
propotion	proportion of patients in each bootstrapping iteration
nfolds	folds to perform cross validation in LASSO
plot_vars	plotting important variables
color	default is steelblue
palette	plotting palette, default is '#999999', using 'RColorBrewer::display.brewer.all()' to see more options
discrete_x	if maximal character length of variables is larger than discrete_x, label will be discrete

Author(s)

Dongqiang Zeng

Examples

res<-best_predictor_binomial(target_data = target, features = features,response = "status",nfolds = 10,permutation = 100)

DongqiangZeng0808/Blasso documentation built on Oct. 12, 2022, 9:44 p.m.