fit_glmnet_with_cv: Wrapper for fitting elastic net model with nested...

In liangyy/mixqtl: Combine total and allele-specific counts for cis-QTL analysis

Description

Build additive linear model to predict y from x using elastic net. It fits the use case when the number of predictors is much larger than the number of observations.

Usage

fit_glmnet_with_cv(
  x,
  y,
  nfold = 5,
  lambda_seq = NULL,
  alpha = 0.5,
  n_times = 3,
  ...
)

Arguments

x	x (N by P, usually P >> N)
y	y (N by 1)
nfold	fold in cross-valiation to pick hyperparameter lambda (default = 5)
lambda_seq	specify the sequence of hyperparameter lambda to search. If don't have such lambda_seq, set it to NULL, and the function will generate one internally (default = NULL)
alpha	alpha parameter in glmnet (defualt = 0.5)
n_times	how many rounds of partitions you'd like to perform to estimate CV MSE for each lambda (it is specific for nested CV procedure; default = 3)
...	additional arguments passed to cv.glmnet

See Also

do_elastic_net for details on nested CV procedure.

liangyy/mixqtl documentation built on Sept. 17, 2020, 11:36 a.m.