otmatch: Statistical Matching using Optimal transport

In RJauslin/transportMatching: Statistical Matching using optimal transport and balanced sampling.

Description

This function computes the statistical matching between two complex survey samples with weighting schemes. The function uses the function transport of the package transport.

Usage

otmatch(
  X1,
  id1,
  X2,
  id2,
  w1,
  w2,
  dist_method = "Euclidean",
  transport_method = "shortsimplex",
  EPS = 1e-09
)

Arguments

X1	A matrix, the matching variables of sample 1.
id1	A character or numeric vector that contains the labels of the units in sample 1.
X2	A matrix, the matching variables of sample 2.
id2	A character or numeric vector that contains the labels of the units in sample 1.
w1	A numeric vector that contains the weights of the sample 1, harmonized by the function harmonize.
w2	A numeric vector that contains the weights of the sample 2, harmonized by the function harmonize.
dist_method	A string that specified the distance used by the function dist of the package proxy. Default Euclidean.
transport_method	A string that specified the distance used by the function transport of the package transport. Default "shortsimplex".

Value

A data.frame that contains the matching. The first two columns contain the unit identities of the two samples. The third column is the final weight. All remaining columns are the matching variables.

Examples

#--- SET UP
N=1000
p=5
X=array(rnorm(N*p),c(N,p))
EPS= 1e-9

n1=100
n2=200

s1=srswor(n1,N)
s2=srswor(n2,N)


id1=(1:N)[s1==1]
id2=(1:N)[s2==1]

d1=rep(N/n1,n1)
d2=rep(N/n2,n2)

X1=X[s1==1,]
X2=X[s2==1,]

#--- HARMONIZATION

re=harmonize(X1,d1,id1,X2,d2,id2)
w1=re$w1
w2=re$w2

#--- STATISTICAL MATCHING WITH OT

object = otmatch(X1,id1,X2,id2,w1,w2)


round(colSums(object$weight*object[,4:ncol(object)]),3)
round(colSums(w1*X1),3)
round(colSums(w2*X2),3)

RJauslin/transportMatching documentation built on Dec. 18, 2021, 8:49 a.m.