CategorModel: R6 class for modeling (fitting and predicting) joint...

In chizhangucb/tmleCommunity: Targeted Maximum Likelihood Estimation for Hierarchical Data

Description

CategorModel inherits from GenericModel class, defining and modeling a conditional density P(A[m]|W,E...) where A[m] is univariate and categorical. By calling self$new(), A[m] will be redefined into number of bins length(levels) (i.e., number of unique categories in A[m]). By calling self$fit(), it fits hazard regressoin Bin_A[m][k] ~ W + E on data (a DatKeepClass class), which is the hazard probaility of the observation of A[m] belongs to bin Bin_A[m][t], given covariates (W, E) and that observation doesn't belong to any precedent bins Bin_A[m][1], Bin_A[m][2], ..., Bin_A[m][k-1].

Usage

CategorModel

Format

An R6Class generator object

Details

• reg - .

• outvar - .

• levels - Numeric vector of all unique categories in outcome outvar.

• nbins - .

• bin_nms - .

Methods

new(reg, DatKeepClass.g0, ...)

Instantiate an new instance of CategorModel for a univariate categorical outcome A[m]

fit(data)

...

predict(newdata)

...

predictAeqa(newdata)

...

Active Bindings

cats

...

See Also

DatKeepClass, RegressionClass, GenericModel, BinaryOutModel

chizhangucb/tmleCommunity documentation built on May 20, 2019, 3:34 p.m.