runCV: Run logit modeling

In novasmedley/gbmSpm: Find predictive sequential patterns for clinical glioblastoma patients

Description

Patients are first split into training and testing partitions. Next, samples with NA features will be removed. Then, training partition is split for cross-validation so that no patient has events in both validation and training. Both partitions are prepped for logit fitting and cross-validation is run.

Usage

runCV(data, cvReps, formula, labelName, lasso = FALSE, llength = NULL,
  lmax = NULL, predictors = NULL, needToRemove = NULL,
  createModelMatrix = FALSE, metric = c("prAUC", "rocAUC"), seed, folds)

Arguments

data	dataframe, rows are samples, cols are features plus some metadata not meant for modeling and will be removed
formula	char or formula object
labelName	char, column name of binary label
lasso	logical, whether to use lasso regularization
llength	num, number of lambdas to consider up to lmax
lmax	num, maximum lambda to consider, cannot be NULL if lambda is NULL
predictors	char, names of columns in data that should be in logit fit data
needToRemove	char, names of columns in data that should not be in logit fit data
createModelMatrix	logical, call model.matrix
metric	char, see aucs
seed	int, seed for split
folds	number of folds

Value

list of logit fits, averaged cross-validation results (if cvReps > 1), and test data partition (one that is formated like training data and is ready to be called by subsequent trained model)

See Also

prepLogitData, getCV, getPerformanceNames

novasmedley/gbmSpm documentation built on May 17, 2019, 10:39 a.m.