DSC_per: Permutation

In hesyserena/DSC: DSC - Distributional Synthetic Controls

Description

Function for evaluating the significance of the estimates using permutation distributions

Usage

DSC_per(
  c_df,
  t_df,
  T0,
  M = 100,
  solver = "SCS",
  ww = 0,
  peridx = 0,
  evgrid = seq(from = 0, to = 1, length.out = 101),
  graph = TRUE,
  y_name = "y",
  x_name = "x"
)

Arguments

c_df	a list of list: each sublist contains observations of every control units at a specific time period. e.g. c_df[[1]][[2]] contains the observations of the second control unit at the first time period.
t_df	a list of observations. e.g. t_df[[1]] contains the observations of the target at the first time period.
T0	an integer specifying the end of pre-treatment period.
M	an integer specifying the number of draws from the uniform distribution for approximating the integral.
solver	a solver for the optimization problem; see installed_solvers in CVXR for more options.
ww	0 or a vector. By default, i.e. 0, arithmetic mean is used for calculating optimal weights. Otherwise, a vector of specific weights is used.
peridx	0 or a vector. By default, i.e. 0, all control units will be used for the permutation. Otherwise, a vector of specific control units will be used.
evgrid	a vector of gridpoints used to evaluate quantile functions.
graph	TRUE/FALSE, indicating whether to output a plot of squared Wasserstein distances.
y_name	a string for the title of the y-axis.
x_name	a string for the title of the x-axis.

Value

DSC_per returns a list containing the following components:

target.dist	a vector of squared Wasserstein distances calculated using original target and control units.
control.dist	a list of squared Wasserstein distances calculated using permutation distributions.

Examples

#simulated data from independent normal distributions
#ex_normal() calls the simulated data
#detail can be found by ??ex_normal
#all control units are used
DSC_per(c_df=ex_normal()$control, t_df=ex_normal()$target, T0=5, M=100, solver="SCS", ww=0, peridx=0, evgrid=seq(from=0, to=1, length.out=1001), graph=TRUE, y_name='y', x_name='x')
#the first, third and fifth control units are used
DSC_per(c_df=ex_normal()$control, t_df=ex_normal()$target, T0=5, M=100, solver="SCS", ww=0, peridx=c(1,3,5), evgrid=seq(from=0, to=1, length.out=1001), graph=TRUE, y_name='y', x_name='x')

hesyserena/DSC documentation built on Dec. 20, 2021, 3:49 p.m.