fit.logreg: Fit logistic regression by using logistic LASSO with lambda=0
In EMLasso: Fit (logistic) LASSO when there is missing data in the predictors

Description Usage Arguments Value Note Author(s) Examples

Fit logistic regression by using logistic LASSO with lambda=0

  fit.logreg(ds, out, wts = rep(1, nrow(ds)),
    verbosity = 0, useCols = NULL, fakeLam,
    imputeDs2FitDsProperties = normalImputationConversion(),
    ..., type.measure = NULL, nfolds = 10, foldid)

  logregLikeGlmnet(x, y, useLambda = Inf, verbosity = 0,
    ...)

`ds`	dataset (`numdfr` or `data.frame`)
`out`	outcome vector
`wts`	weight vector per observation (does not have to sum to 1, and defaults to equal weights)
`verbosity`	The higher this value, the more levels of progress and debug information is displayed (note: in R for Windows, turn off buffered output)
`useCols`	restrict the regression to only include these columns as predictors
`fakeLam`	optionally pass in the lambdas here
`imputeDs2FitDsProperties`	see `imputeDs2FitDs` and `EMLasso`
`...`	passed on to `glmnet`/`glm.fit`. Not allowed: "x", "y", "family", ("weights", "lambda"), "standardize"
`type.measure`	see `cv.glmnet` if `NULL` (the default) no crossvalidation occurs similar to what happens in fit.glmnet
`nfolds`	see `cv.glmnet`
`foldid`	see `cv.glmnet`
`x`	model matrix (as supported by `glm`)
`y`	outcomes (as supported by `glm`)
`useLambda`	lambda item of the return value

glmnet object

similar to the return value of glmnet

The warning in the old function pointed me to the fact that the weights are _not_ probability weights in a binomial glm!! So, as an alternative, we use glmnet with lambda=0 !!!

Nick Sabbe nick.sabbe@ugent.be