Fixes some existing issues from myself and Nima, and completely overhauls the following:
How graphs (and other attributes) are stored in a CausalTable; now arbitrary fixed variables in the DGP can be stored in an "arrays" attribute, including adjacency matrices of graphs
How random data is generated; introduces special character | to indicate summary functions in the DGP, and allows arbitrary code to be executed in between draws of random variables using =
How summary functions are implemented; removed old summaries to make way for more general ones, i.e. arbitrary Sum options which simply multiply by an arbitrary matrix from the DGP
Other changes:
Changes old DataGeneratingProcess to StructuralCausalModel; now the new DataGeneratingProcess stores just the infrastructure needed to draw random variables, while StructuralCausalModel stores information about causal variables labels (treatment, response, etc.) in order to randomly draw CausalTables
Some minor changes with results of random draws from DGPs
Fixes some existing issues from myself and Nima, and completely overhauls the following:
|
to indicate summary functions in the DGP, and allows arbitrary code to be executed in between draws of random variables using=
Sum
options which simply multiply by an arbitrary matrix from the DGPOther changes: