This package implements methods for maximum entropy synthetic controls using multiple outcomes.
To install this package, first ensure that devtools
is installed with
install.packages("devtools")
then install the package from GitHub with
devtools::install_github("ebenmichael/ents")
The following requires a data frame outcomes
which at a minimum has the following columns:
outcome_id
: ID or name of the outcometime
: Time that outcome was measuredoutcome
: Value of the outcometreated
: Whether the unit is treated
and a data frame metadata
which at a minimum has the following columns:
t_int
: Time of intervention/treatment
Four methods of imputing a synthetic control are implemented:
get_synth
: The Abadie, Diamond, Hainmueller (2010) synthetic controls estimator, fit usingSynth
get_l2_entropy
: The maximum entropy synthetic controls estimatorget_dr
: The maximum entropy synthetic controls estimator augmented with a linear outcome modelget_ipw
: An IPW estimator fit with regularized logistic regression
Each function takes in outcomes
and metadata
along with hyper-parameters, and returns as list including a dataframe with the synthetic control added and the synthetic control weights.