SL.gilleskie: Adaptive Hazard method of Gilleskie and Mroz

In wuziyueemory/twostageSL: Two-Stage Super Learner Prediction Function

Description

This function implements the estimator of Gilleskie and Mroz (2004), which estimates the conditional mean costs by adaptively modeling the conditional hazard function and back-transforming into an estimate of the conditional mean. For more description, we refer users to the original paper.

Usage

SL.gilleskie(
  Y,
  X,
  newX,
  family = gaussian(),
  obsWeights = rep(1, length(Y)),
  kValues = c(5, 15, 25),
  maxPoly = 2,
  pValThresh = 0.05,
  ...
)

Arguments

Y	A numeric outcome variable
X	A data.frame of covariates constituting the training sample
newX	A data.frame with the same column names and format as X constituting the validation sample.
family	Gaussian only
obsWeights	Observation-level weights (not currently used)
kValues	Number of intervals to bin the variable into in order to estimate the discrete hazard function
maxPoly	The largest degree polynomial to be used in fitting the discrete hazard function
pValThresh	The threshold for p-values used in the covariate selection process
...	Other arguments (not currently used)

Value

pred

Predicted outcomes based on predictors in newX

fit

A list with named entries object (the fitted hazard regression object), maxK (the selected number of partitions), and hK (the mean outcome in each partition)

References

Gilleskie DB, Mroz TA (2004). “A flexible approach for estimating the effects of covariates on health expenditures.” Journal of Health Economics, 23(2), 391–418.

Examples

# load cost data
data(cost_data)

fit_gilleskie <- SL.gilleskie(Y = cost_data$totalcost, X = cost_data[, c("female", "race")],
                              newX = cost_data[, c("female", "race")])

wuziyueemory/twostageSL documentation built on Oct. 19, 2020, 3:45 p.m.