fit_prog_reg: Use a separate regularized regression for each post period to...

In ebenmichael/ents: Maximum Entropy Synthetic Controls

Description

Use a separate regularized regression for each post period to fit E[Y(0)|X]

Usage

fit_prog_reg(X, y, trt, alpha = 1, lambda = NULL, poly_order = 1,
  type = "sep")

Arguments

X	Matrix of covariates/lagged outcomes
y	Matrix of post-period outcomes
trt	Vector of treatment indicator
alpha	Mixing between L1 and L2, default: 1 (LASSO)
lambda	Regularization hyperparameter, if null then CV
poly_order	Order of polynomial to fit, default 1
type	How to fit outcome model(s) sep Separate outcome models avg Average responses into 1 outcome multi Use multi response regression in glmnet

Value

• y0hat Predicted outcome under control

• params Regression parameters

ebenmichael/ents documentation built on May 31, 2019, 8:45 p.m.