tests/testthat/test-precision.R

In rcbalance: Large, Sparse Optimal Matching with Refined Covariate Balance

library(rcbalance)
if (requireNamespace("optmatch", quietly = TRUE)){
  library(optmatch)
context('Precision for distances and penalties')

#skip_on_cran()

data(nuclearplants)
#reorder nuclearplants dataframe so treated units come first
nuclearplants <- nuclearplants[order(nuclearplants$pr, decreasing = TRUE),]

my.dist.struct <- build.dist.struct(z = nuclearplants$pr, 
	X = subset(nuclearplants[c('date', 't1', 't2', 'cap','cum.n')]), 
	exact = nuclearplants$ne, calip.option = 'none')
	
match.out <- rcbalance(my.dist.struct, treated.info = nuclearplants[1:10,], control.info = nuclearplants[11:32,], fb.list = list('ct',c('ct','bw')))	

test_that('Fine balance is achieved', {
	expect_equal(match.out$fb.tables[[1]][,1], match.out$fb.tables[[1]][,2])
	expect_equal(match.out$fb.tables[[2]][,1], match.out$fb.tables[[2]][,2])
})

#THIS CHECK NOW DEPRECATED, NOT TOO IMPORTANT AND HARD TO MAKE CONSISTENT
# BOTH LOCALLY AND ON CRAN.
#test_that('Overly large distances are recognized and caught', {
#	new.dist <- matrix(c(1:3,5)*1e-5, nrow=2, ncol = 2) + .Machine$integer.max - 2
	#expect_warning(expect_error(rcbalance(new.dist), 
	# 'Integer overflow in penalties!  Run with a higher tolerance, a lower penalty value, or fewer levels of fine balance.'),
	# 'NAs introduced by coercion to integer range')
	#new.match <- rcbalance(new.dist)
	#expect_equal(as.vector(new.match$matches), c(1:2))
#})
}