BalancedSampling (version 1.4)

flightphase: Flight phase of the cube method

Description

This is a fast implementation of the flight phase of the cube method. To have a fixed sample size, include the inclusion probabilities as a balancing variable in Xbal and make sure the inclusion probabilities sum to a positive integer.

Usage

flightphase(prob,Xbal)

Arguments

prob

vector of length N with inclusion probabilities

Xbal

matrix of balancing auxiliary variables of N rows and q columns

Value

• Returns a vector of length N with new probabilities, where at most q are non-integer.

References

Deville, J. C. and Tillé, Y. (2004). Efficient balanced sampling: the cube method. Biometrika, 91(4), 893-912.

Chauvet, G. and Tillé, Y. (2006). A fast algorithm for balanced sampling. Computational Statistics, 21(1), 53-62.

Examples

# Example 1
# Select sample and check balance
set.seed(12345);
N = 100; # population size
n = 10; # sample size
p = rep(n/N,N); # inclusion probabilities
X = cbind(p,runif(N),runif(N)); # matrix of auxiliary variables

pflight = flightphase(p,X);

# check balance after flight
colSums(X)
colSums(X/p*pflight)

# select final sample as indicators
indicators = landingphase(p,pflight,X);

# check final balance
colSums(X)
colSums(X/p*indicators)

# final sample as indexes
s = (1:N)[indicators==1];