tmle.Sl.dbarts2 function

Super Learner wrappers for modeling and prediction using bart in the dbarts package

These functions are used internally, not typically called by the user

tmle.SL.dbarts2(Y, X, newX, family, obsWeights, id, sigest = NA, sigdf = 3, 
        sigquant = 0.90, k = 2, power = 2.0, base = 0.95, binaryOffset = 0.0, 
        ntree = 200, ndpost = 1000, nskip = 100, printevery = 100,  keepevery = 1,  
        keeptrainfits = TRUE, usequants = FALSE, numcut = 100,printcutoffs = 0,  
        nthread = 1,   keepcall = TRUE,verbose = FALSE, ...)
tmle.SL.dbarts.k.5(Y, X, newX, family, obsWeights, id, sigest = NA, sigdf = 3, 
        sigquant = 0.90, k = 0.5, power = 2.0, base = 0.95, binaryOffset = 0.0, 
        ntree = 200, ndpost = 1000, nskip = 100, printevery = 100,  keepevery = 1,  
        keeptrainfits = TRUE, usequants = FALSE, numcut = 100,printcutoffs = 0,  
        nthread = 1,   keepcall = TRUE,verbose = FALSE, ...)
## S3 method for class 'tmle.SL.dbarts2'
predict(object, newdata, family, ...)

Arguments

• Y: Dependent variable
• X: Predictor covariate matrix or data frame used as training set
• newX: Predictor covariate matrix or data frame for which predictions should be made
• family: Regression family, 'gaussian' or 'binomial'
• obsWeights: observation-level weights
• id: identifier to group observations, not used
• sigest: An estimate of error variance. See bart documentation
• sigdf: Degrees of freedom for error variance prior. See bart documentation
• sigquant: Quantile of error variance prior. See bart documentation
• k: Tuning parameter that controls smoothing. Larger values are more conservative, see Details
• power: Power parameter for tree prior
• base: Base parameter for tree prior
• binaryOffset: Allows fits with probabilities shrunk towards values other than 0.5. See bart documentation
• ntree: Number of trees in the sum-of-trees formulation
• ndpost: Number of posterior draws after burn in
• nskip: Number of MCMC iterations treated as burn in
• printevery: How often to print messages
• keepevery: Every keepevery draw is kept to be returned to the user
• keeptrainfits: If TRUE the draws of $f(x)$ for $x$ corresponding to the rows of x.train are returned
• usequants: Controls how tree decisions rules are determined. See bart documentation
• numcut: Maximum number of possible values used in decision rules
• printcutoffs: Number of cutoff rules to print to screen. $0$ prints nothing
• nthread: Integer specifying how many threads to use
• keepcall: Returns the call to BART when TRUE
• verbose: Ignored for now
• ``: Additional arguments passed on to plot or control functions
• object: Object of type tmle.SL.dbarts2
• newdata: Matrix or dataframe used to get predictions from the fitted model

Returns

an object of type tmle.SL.dbarts2 used internally by Super Learner

Details

tmle.SL.dbarts2 is in the default library for estimating $Q$. It uses the default setting in the dbarts package, $k=2$. tmle.SL.dbarts.k.5 is used to estimate the components of $g$. It sets $k=0.5$, to avoid shrinking predicted values too far from $(0,1)$. See bart documentation for more information.

Author(s)

Chris Kennedy and Susan Gruber

See Also

SuperLearner