SuperLearner.hazard: Hazard estimation using binary Super Learner
In ecpolley/SuperLearner_Old: Super Learner Prediction

Description Usage Arguments Value Author(s) See Also

Hazard estimation using binary Super Learner

SuperLearner.hazard(N.delta, n.controls = 1, time.df = 5, time, X, newX, time.newX, SL.library, V = 20, shuffle = FALSE, verbose = FALSE, family = binomial(), method="NNLS", id = NULL, save.fit.library = TRUE, trim.logit=0.0001, discreteTime = TRUE, obsWeights=NULL)

`N.delta`	Discrete survival process. often created by `create.discrete`
`n.controls`	number of randomly sampled observations from risk set using `incidence.sample`. Set to 0 for entire sample.
`time.df`	degrees of freedom for the `gam` model. Currently not implemented
`time`	the time variable. often `delta.u` from `create.discrete`
`X`	the baseline variables. Do not include the outcome nor time
`newX`	test data frame
`time.newX`	time points in the newX data frame
`SL.library`	algorithms for binary Super Learner
`V`	number of cross validation folds for the Super Learner
`shuffle`	shuffle rows in data frame before creating cross validation folds
`verbose`	more detailed output
`family`	binomial
`discreteTime`	currently not implemented
`method`	Loss function for combining prediction in the library. Currently either "NNLS" (the default) or "NNloglik". NNLS is non-negative least squares and will work for both gaussian and binomial. NNloglik is a non-negative binomial likelihood maximization
`id`	subject identification variable
`save.fit.library`	logical variable for saving the fit of each algorithm in `SL.library`
`trim.logit`	Only used if `method="NNloglik"`. specifies a truncation level for the logit function for stability.
`obsWeights`	observation weights

`SL.predict`	predicted probability of having an event
`cand.names`	gives a list of all algorithms in the library, including any screening algorithms
`SL.library`	gives a list of all algorithms in the library
`init.coef`	coefficient estimates from the non-negative least squares
`coef`	Coefficient estimates in the super learner
`library.predict`	predicted values from all candidates in `SL.library` estimated on the entire training sample
`cv.risk`	V-fold cross-validated risk estimates for each algorithm in the library
`newZ`	predicted values from all candidates in `SL.library` estimated in the V-fold cross validation
`fit.library`	a list containing the fit of each model in SL.library on the full data set
`id`	cluster identification variable
`namesX`	The variable names in the X data frame
`DATA.split`	a list with the cross-validation splits
`method`	method used to estimate weight for prediction in library
`whichScreen`	a logical matrix indicating which variables were selected for each screening algorithm
`trim.logit`	Only used if `method="NNloglik"`. specifies a truncation level for the logit function for stability.
`errorsInCVLibrary`	A logical vector with length equal to the number of algorithms in the library. equal to 1 if the corresponding algorithm failed on any of the cross-validation splits
`errorsInLibrary`	A logical vector with length equal to the number of algorithms in the library. equal to 1 if the corresponding algorithm failed on full data