use_nested_cvglmnet.Rd

% Generated by roxygen2: do not edit by hand
% Please edit documentation in R/use_nested_cvglmnet.R
\name{use_nested_cvglmnet}
\alias{use_nested_cvglmnet}
\title{Use nested CV with Grid Search}
\usage{
use_nested_cvglmnet(
  data = NULL,
  response,
  model_matrix = NULL,
  shorten_names = F,
  verbose = F,
  type_measure = "class",
  alpha = 0.5,
  make_model = F,
  variables = NULL,
  stratify = T,
  repeats = 1,
  k_outer_loop = 10,
  k_inner_loop = 5,
  seed = 12,
  standardize = F,
  ...
)
}
\arguments{
\item{response}{Respose vector for glmnet or whole classification matrix from assay if predicted_prob is a list.}

\item{shorten_names}{Logical - Shorten rownames of model matrix, might be usefull when working with tcga data}

\item{...}{Additional args for glmnet}

\item{predicted_prob}{Predicted probabilities from make_prediction or any other fit as list or single vector}

\item{prediction}{Named vector of prediction.}
}
\description{
This function makes a data frame with coefficient from glmnet
}
\details{
Able to do stratified cv. Be aware if binary classification is very irregular, like a 95/5 split, then a 10 fold cross validation might fail, if you only have 100 samples.
Standardize = F in glmnet because supplied model matrix should already be standardized.
}