GenericModel: Generic R6 class for modeling (fitting and predicting) P(A=a|W=w) where A can be a multivariate (A[1], ..., A[k]) and each A[i] can be binary, categorical or continous
Description
This R6 class Class for defining, fitting and predicting the probability model
P(A|W) under g_star or under g_0 for variables
(A,W). Defines and manages the factorization of the multivariate conditional
probability model P(A=a|...) into univariate regression models
A[j] ~ A[j-1] + ... + A[1] + W. The class self$new method automatically
figures out the correct joint probability factorization into univariate conditional
probabilities based on name ordering provided by (A_nms, W_nms).
When the outcome variable A[j] is binary, this class will automatically call
a new instance of BinaryOutcomeModel class.
Provide self$fit() function argument data as a DataStorageClass class object.
This data will be used for fitting the model P(A|W).
Provide self$fit() function argument newdata (also as DataStorageClass class) for predictions of the type
P(A=1|W=w), where w values are coming from newdata object.
Finally, provide self$predictAeqa function newdata argument
(also DataStorageClass class object) for getting the likelihood predictions P(A=sa|W=w), where
both, sa and sw values are coming from newdata object.