Learn R Programming
DynTxRegime (version 3.01)

.newQLearn: A Step of the Q-Learning Algorithm

Description

Regression of a single stage. For multiple stage analyses, function called for each decision point. Method is not exported.

Usage

.newQLearn(moMain, moCont, fSet, response, ...)
# S4 method for modelObj,modelObj,function,vector
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)
# S4 method for modelObj,NULL,function,vector
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)
# S4 method for NULL,modelObj,function,vector
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)
# S4 method for modelObj,modelObj,NULL,vector
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)
# S4 method for modelObj,NULL,NULL,vector
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)
# S4 method for NULL,modelObj,NULL,vector
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)
# S4 method for ModelObj_SubsetList,ModelObj_SubsetList,function,vector
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)
# S4 method for ModelObj_SubsetList,NULL,function,vector
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)
# S4 method for NULL,ModelObj_SubsetList,function,vector
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)
# S4 method for modelObj,modelObj,function,QLearn
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)
# S4 method for modelObj,NULL,function,QLearn
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)
# S4 method for NULL,modelObj,function,QLearn
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)
# S4 method for modelObj,modelObj,NULL,QLearn
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)
# S4 method for modelObj,NULL,NULL,QLearn
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)
# S4 method for NULL,modelObj,NULL,QLearn
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)
# S4 method for ModelObj_SubsetList,ModelObj_SubsetList,function,QLearn
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)
# S4 method for ModelObj_SubsetList,NULL,function,QLearn
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)
# S4 method for NULL,ModelObj_SubsetList,function,QLearn
.newQLearn(moMain, moCont, fSet, response, data, txName, iter, suppress)

Arguments

moMain
"modelObj"(s) for main effects component of outcome.
moCont
"modelObj"(s) for contrasts component of outcome.
fSet
"function" defining feasible treatments and/or modeling subsets.
response
If vector, outcome of interest. If "QLearn", value object returned by previous call to Q-Learning algorithm
...
Used to pass arguments that are required but the class of which does not determine the method that is selected.
data
"data.frame" of covariates and treatment history.
txName
"character" indicating column header of data for treatment variable.
iter
Maximum number of iterations if iterative algorithm requested.
suppress
T/F indicating if screen prints are generated.

Examples

Run this code

data(bmiData)

y <- -(bmiData$month12BMI - bmiData$baselineBMI) / bmiData$baselineBMI * 100

bmiData$A2 <- as.factor(bmiData$A2)
bmiData$A1 <- as.factor(bmiData$A1)

moMain <- buildModelObj(model = ~parentBMI+baselineBMI+month4BMI, 
                        solver.method = 'lm')
moCont <- buildModelObj(model = ~parentBMI+baselineBMI+month4BMI, 
                        solver.method = 'lm')

obj1 <- DynTxRegime:::.newQLearn(moMain = moMain, 
                                 moCont = moCont, 
                                 fSet = NULL, 
                                 response = y,
                                 data = bmiData, 
                                 txName = 'A2', 
                                 iter = 0L, 
                                 suppress = TRUE)

is(obj1)

coef(obj1)
DTRstep(obj1)
estimator(obj1)
fitObject(obj1)
optTx(obj1)
optTx(obj1,bmiData)
outcome(obj1)
plot(obj1)
propen(obj1)
show(obj1)
summary(obj1)

moMain2 <- buildModelObj(model = ~parentBMI+baselineBMI, 
                         solver.method = 'lm')
moCont2 <- buildModelObj(model = ~parentBMI+baselineBMI, 
                         solver.method = 'lm')

obj2 <- DynTxRegime:::.newQLearn(moMain = moMain2, 
                                 moCont = moCont2,  
                                 fSet = NULL,  
                                 response = obj1,
                                 data = bmiData,  
                                 txName = 'A1',  
                                 iter = 0L,  
                                 suppress = TRUE)

is(obj2)

coef(obj2)
DTRstep(obj2)
estimator(obj2)
fitObject(obj2)
optTx(obj2)
optTx(obj2,bmiData)
outcome(obj2)
plot(obj2)
propen(obj2)
show(obj2)
summary(obj2)

fSet <- function(data){
          subsets <- list(list("subset1",c("CD","MR")),
                          list("subset2",c("CD","MR")))
          txOpts <- character(nrow(data))
          txOpts[data$baselineBMI < 32] <- "subset2"
          txOpts[data$baselineBMI >=32] <- "subset1"
          return(list("subsets" = subsets, "txOpts" = txOpts))
        }

moMain3 <- list()
moMain3[[1L]] <- buildModelObjSubset(model = ~parentBMI+baselineBMI, 
                                     solver.method = 'lm',  
                                     subset = "subset1")
moMain3[[2L]] <- buildModelObjSubset(model = ~parentBMI+baselineBMI,  
                                     solver.method = 'lm',  
                                     subset = "subset2")
moCont3 <- list()
moCont3[[1L]] <- buildModelObjSubset(model = ~parentBMI+baselineBMI,  
                                     solver.method = 'lm',  
                                     subset = "subset1")
moCont3[[2L]] <- buildModelObjSubset(model = ~parentBMI+baselineBMI,  
                                     solver.method = 'lm',  
                                     subset = "subset2")

moMain3 <- DynTxRegime:::.newModelObjSubset(moMain3)
moCont3 <- DynTxRegime:::.newModelObjSubset(moCont3)

obj2 <- DynTxRegime:::.newQLearn(moMain = moMain3, 
                                 moCont = moCont3, 
                                 fSet = fSet, 
                                 response = obj1,
                                 data = bmiData, 
                                 txName = 'A1', 
                                 iter = 0L, 
                                 suppress = TRUE)

is(obj2)

coef(obj2)
DTRstep(obj2)
estimator(obj2)
fitObject(obj2)
optTx(obj2)
optTx(obj2,bmiData)
outcome(obj2)
plot(obj2)
propen(obj2)
show(obj2)
summary(obj2)

Run the code above in your browser using DataLab