Roadmap for Econometrics.jl

[Nosferican]

I wanted to get some feedback and thoughts on how best port the information I give for the estimators in Econometrics.jl (see examples). These models cover the ones supported by FixedEffectsModels (e.g., instrumental variables, absorption of features, robust variance covariance estimators) as well as some of the nonlinear models in GLM.jl (e.g., discrete choice models). Would it be best to work on implementing the API?

[jmboehm]

I think a TableRegressionModel is a good interface for nonlinear model estimates. If you want to show more statistics, we can probably add them in the footer of the table through the regression_statistics option.

[Nosferican]

Any reason why the API should be at TableRegressionModel?

struct TableRegressionModel{M,T} <: RegressionModel
    model::M
    mf::ModelFrame
    mm::ModelMatrix{T}
end

Those structs don't play that nicely with instrumental variables, not a single ModelFrame/ModelMatrix which are legacy as the new recommended method is modelcols.
Is there any advantage to that over the CoefTable which is part of the API? TableRegressionModel is at StatsModels and technically not part of the abstraction for RegressionModel?