Someone noted to me that FixedEffectModels.jl is tricky to use AD on because there are so many explicit Float64 type constraints -- does anyone have a good sense of how much effort it would take to remove/parameterize/reduce the explicit type constraints here?
As an example, the FixedEffectModel struct has a lot of Float64 explicit types that could be parametric instead.
|
struct FixedEffectModel <: RegressionModel |
|
coef::Vector{Float64} # Vector of coefficients |
|
vcov::Matrix{Float64} # Covariance matrix |
|
vcov_type::CovarianceEstimator |
|
nclusters::Union{NamedTuple, Nothing} |
|
|
|
esample::BitVector # Is the row of the original dataframe part of the estimation sample? |
|
residuals::Union{AbstractVector, Nothing} |
|
fe::DataFrame |
|
fekeys::Vector{Symbol} |
|
|
|
|
|
coefnames::Vector # Name of coefficients |
|
yname::Union{String, Symbol} # Name of dependent variable |
|
formula::FormulaTerm # Original formula |
|
formula_predict::FormulaTerm |
|
contrasts::Dict |
|
|
|
nobs::Int64 # Number of observations |
|
dof_residual::Int64 # nobs - degrees of freedoms |
|
|
|
rss::Float64 # Sum of squared residuals |
|
tss::Float64 # Total sum of squares |
|
r2::Float64 # R squared |
|
adjr2::Float64 # R squared adjusted |
|
|
|
F::Float64 # F statistics |
|
p::Float64 # p value for the F statistics |
|
|
|
# for FE |
|
iterations::Union{Int, Nothing} # Number of iterations |
|
converged::Union{Bool, Nothing} # Has the demeaning algorithm converged? |
|
r2_within::Union{Float64, Nothing} # within r2 (with fixed effect |
|
|
|
# for IV |
|
F_kp::Union{Float64, Nothing} # First Stage F statistics KP |
|
p_kp::Union{Float64, Nothing} # First Stage p value KP |
|
end |
Is there an appetite for this? I think it'd be lovely to be able to AD through high-dimensional fixed effect estimates.
