fit_glmnet: Fit elastic net, LASSO or ridge regression model
In emil: Evaluation of Modeling without Information Leakage

Description Usage Arguments Details Value Author(s) References See Also

Using the glmnet package implementation.

fit_glmnet(x, y, family, nfolds, foldid, alpha = 1, lambda = NULL, ...)

fit_ridge_regression(...)

fit_lasso(...)

`x`	Dataset.
`y`	Response vector. Can be of many different types for solving different problems, see `glmnet`.
`family`	Determines the the type of problem to solve. Auto detected if `y` is numeric or survival. See `family` for details.
`nfolds`	See `cv.glmnet`.
`foldid`	See `cv.glmnet`.
`alpha`	Regularization parameter, see `glmnet`.
`lambda`	Regularization parameter, see `glmnet`.
`...`	Sent to `fit_glmnet` or `cv.glmnet`.

The alpha parameter of glmnet controls the type of penalty. Use 0 (default) for lasso only, 1 for ridge only, or an intermediate for a combination. This is typically the parameter to tune on. The shrinkage, controlled by the lambda parameter, can be left unspecified for internal tuning (works the same way as fit_glmnet).

Fitted elastic net model.

Christofer Bäcklin

Friedman J, Hastie T, Tibshirani R (2010). Regularization Paths for Generalized Linear Models via Coordinate Descent. Journal of Statistical Software, 33(1), 1–22. doi:10.18637/jss.v033.i01.

emil, predict_glmnet, importance_glmnet, modeling_procedure

emil documentation built on Aug. 1, 2018, 1:03 a.m.