DynTxRegime (version 3.2)

owl: Outcome Weighted Learning

Description

Estimation of an optimal treatment regime using outcome weighted learning. The convex surrogate of the 0-1 loss function is taken to be the hinge function. The method is for single-decision-point analysis with binary treatment options.

Usage

owl(..., moPropen, data, reward, txName, regime, lambdas = 2.0, cvFolds = 0L, 
    kernel = "linear", kparam = NULL, verbose = TRUE)

Arguments

…

ignored. Included to require named input.

moPropen

An object of class "modelObj." This object specifies the model of the propensity for treatment regression and the R methods to be used to obtain parameter estimates and predictions. The method specified to obtain predictions must return the prediction on the scale of the probability, i.e., returned values must be in the interval (0,1). See moPropen for further information.

data

An object of class "data.frame." The covariates and treatment histories.

reward

An object of class "vector." A vector of the outcome of interest.

txName

An object of class "character." The column header of the stage treatment variable as given in input data. Treatment must be binary and will be recoded as -1/+1 if not provided as such.

regime

An object of class "formula." The formula defines the decision rule, i.e., the covariates to be included in the kernel.

lambdas

An object of class "numeric." Tuning parameter to avoid overfitting. If more than one is given and cvFolds > 0, cross-validation will be used to select an optimal from among those provided.

cvFolds

An object of class "integer." If cross-validation is to be used to find an optimal lambda and/or kernel parameter, the number of folds to use in the CV.

kernel

An object of class "character." In conjunction with input kparam, specification of the kernel function to be used. Must be one of {'linear', 'poly', or 'radial'}. If 'linear,' the linear kernel; kparam is ignored. If 'poly,' the polynomial kernel; kparam must be specified. If 'radial,' the Gaussian radial basis function kernel; kparam must be specified.

kparam

An object of class "numeric." If input kernel = 'linear', this input is ignored. If input kernel = 'poly', this input is the order of the polynomial. If input kernel = 'radial', this input is sigma; i.e., $$K(x,y) = exp(||x-y||^2 / (2*sigma^2)).$$ For kernel = 'radial', a vector of kernel parameters can be provided and cross-validation will be used to determine the optimal of those provided. Note that input cvFolds must be > 0.

verbose

An object of class "logical.". If FALSE, screen prints will be suppressed.

Value

Returns an object of class "OWL" that inherits directly from class "DynTxRegime."

Methods

coef

signature(object = "OWL"): Retrieve parameter estimates for all regression steps.

cvInfo

signature(object = "OWL"): Retrieve cross-validation results.

DTRstep

signature(object = "OWL"): Retrieve description of method used to create object.

estimator

signature(x = "OWL"): Retrieve the estimated value of the estimated optimal regime for the training data set.

fitObject

signature(object = "OWL"): Retrieve value object returned by regression methods.

optimObj

signature(object = "OWL"): Retrieve value object returned by optimization routine.

optTx

signature(x = "OWL", newdata = "missing"): Retrieve the estimated optimal treatment regime for training data set.

optTx

signature(x = "OWL", newdata = "data.frame"): Estimate the optimal treatment regime for newdata.

plot

signature(x = "OWL"): Generate plots for regression analyses.

print

signature(object = "OWL"): Print main results of analysis.

propen

signature(x = "OWL"): Retrieve value object returned by propensity score regression methods.

regimeCoef

signature(object = "OWL"): Retrieve the estimated decision function parameter estimates.

show

signature(object = "OWL"): Show main results of analysis.

summary

signature(object = "OWL"): Retrieve summary information from regression analyses.

References

Zhao, Y-Q., Zeng, D., Rush, A. J., and Kosrok, M. R. (2012). Estimating Individualized Treatment Rules Using Outcome Weighted Learning. Journal of the American Statistical Association, 107, 1106--1118.

Examples

# NOT RUN {
# Load and process data set
  data(bmiData)

  # define response y to be the negative 12 month
  # change in BMI from baseline
  bmiData$y <- -100*(bmiData[,6] - bmiData[,4])/bmiData[,4]


# Constant propensity model
  moPropen <- buildModelObj(model = ~1,
                            solver.method = 'glm',
                            solver.args = list('family'='binomial'),
                            predict.method = 'predict.glm',
                            predict.args = list(type='response'))

  owlRes <- owl(moPropen = moPropen,
                data = bmiData, reward = bmiData$y,  txName = "A2", 
                regime = ~ parentBMI + month4BMI)

##Available methods

  # Coefficients of the propensity score regression
  coef(owlRes)

  # Description of method used to obtain object
  DTRstep(owlRes)

  # Estimated value of the optimal treatment regime for training set
  estimator(owlRes)

  # Value object returned by propensity score regression method
  fitObject(owlRes)

  # Summary of optimization routine
  optimObj(owlRes)

  # Estimated optimal treatment for training data
  optTx(owlRes)

  # Estimated optimal treatment for new data
  optTx(owlRes, bmiData)

  # Plots if defined by propensity regression method
  dev.new()
  par(mfrow = c(2,4))

  plot(owlRes)
  plot(owlRes, suppress = TRUE)

  # Value object returned by propensity score regression method
  propen(owlRes)

  # Parameter estimates for decision function
  regimeCoef(owlRes)

  # Show main results of method
  show(owlRes)

  # Show summary results of method
  summary(owlRes)
 
# }