[epclust.git] / epclust / tests / testthat / test.clustering.R

context("clustering")

test_that("computeSynchrones behave as expected",
{
	n = 300
	x = seq(0,9.5,0.1)
	L = length(x) #96 1/4h
	K = 3
	s1 = cos(x)
	s2 = sin(x)
	s3 = c( s1[1:(L%/%2)] , s2[(L%/%2+1):L] )
	#sum((s1-s2)^2) == 96
	#sum((s1-s3)^2) == 58
	#sum((s2-s3)^2) == 38
	s = list(s1, s2, s3)
	series = matrix(nrow=L, ncol=n)
	for (i in seq_len(n))
		series[,i] = s[[I(i,K)]] + rnorm(L,sd=0.01)
	getRefSeries = function(indices) {
		indices = indices[indices <= n]
		if (length(indices)>0) series[,indices] else NULL
	}
	synchrones = computeSynchrones(bigmemory::as.big.matrix(cbind(s1,s2,s3)), getRefSeries,
		n, 100, sync_mean=TRUE, verbose=TRUE, parll=FALSE)

	expect_equal(dim(synchrones), c(L,K))
	for (i in 1:K)
		expect_equal(synchrones[,i], s[[i]], tolerance=0.01)
})

# Helper function to divide indices into balanced sets
test_that("Helper function to spread indices work properly",
{
	indices <- 1:400

	# bigger nb_per_set than length(indices)
	expect_equal(epclust:::.spreadIndices(indices,500), list(indices))

	# nb_per_set == length(indices)
	expect_equal(epclust:::.spreadIndices(indices,400), list(indices))

	# length(indices) %% nb_per_set == 0
	expect_equal(epclust:::.spreadIndices(indices,200),
		c( list(indices[1:200]), list(indices[201:400]) ))
	expect_equal(epclust:::.spreadIndices(indices,100),
		c( list(indices[1:100]), list(indices[101:200]),
			list(indices[201:300]), list(indices[301:400]) ))

	# length(indices) / nb_per_set == 1, length(indices) %% nb_per_set == 100
	expect_equal(epclust:::.spreadIndices(indices,300), list(indices))
	# length(indices) / nb_per_set == 2, length(indices) %% nb_per_set == 42
	repartition <- epclust:::.spreadIndices(indices,179)
	expect_equal(length(repartition), 2)
	expect_equal(length(repartition[[1]]), 179 + 21)
	expect_equal(length(repartition[[1]]), 179 + 21)
})

test_that("clusteringTask1 behave as expected",
{
	n = 900
	x = seq(0,9.5,0.1)
	L = length(x) #96 1/4h
	K1 = 60
	s = lapply( seq_len(K1), function(i) x^(1+i/30)*cos(x+i) )
	series = matrix(nrow=L, ncol=n)
	for (i in seq_len(n))
		series[,i] = s[[I(i,K1)]] + rnorm(L,sd=0.01)
	getSeries = function(indices) {
		indices = indices[indices <= n]
		if (length(indices)>0) series[,indices] else NULL
	}
	wf = "haar"
	ctype = "absolute"
	getContribs = function(indices) curvesToContribs(series[,indices],wf,ctype)
	require("cluster", quietly=TRUE)
	browser()
	algoClust1 = function(contribs,K) cluster::pam(contribs,K,diss=FALSE)$id.med
	indices1 = clusteringTask1(1:n, getContribs, K1, algoClust1, 75, verbose=TRUE, parll=FALSE)
	medoids_K1 = getSeries(indices1)

	expect_equal(dim(medoids_K1), c(L,K1))
	# Not easy to evaluate result: at least we expect it to be better than random selection of
	# medoids within initial series
	distorGood = computeDistortion(series, medoids_K1)
	for (i in 1:3)
		expect_lte( distorGood, computeDistortion(series,series[,sample(1:n, K1)]) )
})

test_that("clusteringTask2 behave as expected",
{
	n = 900
	x = seq(0,9.5,0.1)
	L = length(x) #96 1/4h
	K1 = 60
	K2 = 3
	#for (i in 1:60) {plot(x^(1+i/30)*cos(x+i),type="l",col=i,ylim=c(-50,50)); par(new=TRUE)}
	s = lapply( seq_len(K1), function(i) x^(1+i/30)*cos(x+i) )
	series = matrix(nrow=L, ncol=n)
	for (i in seq_len(n))
		series[i,] = s[[I(i,K1)]] + rnorm(L,sd=0.01)
	getRefSeries = function(indices) {
		indices = indices[indices <= n]
		if (length(indices)>0) series[,indices] else NULL
	}
	# Artificially simulate 60 medoids - perfect situation, all equal to one of the refs
	medoids_K1 = bigmemory::as.big.matrix( sapply( 1:K1, function(i) s[[I(i,K1)]] ) )
	medoids_K2 = clusteringTask2(medoids_K1, K2, getRefSeries, n, 75, verbose=TRUE, parll=FALSE)

	expect_equal(dim(medoids_K2), c(L,K2))
	# Not easy to evaluate result: at least we expect it to be better than random selection of
	# medoids within 1...K1 (among references)
	distorGood = computeDistortion(series, medoids_K2)
	for (i in 1:3)
		expect_lte( distorGood, computeDistortion(series,medoids_K1[,sample(1:K1, K2)]) )
})
Commit	Line	Data
	1	context("clustering")
	2
	3	test_that("computeSynchrones behave as expected",
	4	{
	5	n = 300
	6	x = seq(0,9.5,0.1)
	7	L = length(x) #96 1/4h
	8	K = 3
	9	s1 = cos(x)
	10	s2 = sin(x)
	11	s3 = c( s1[1:(L%/%2)] , s2[(L%/%2+1):L] )
	12	#sum((s1-s2)^2) == 96
	13	#sum((s1-s3)^2) == 58
	14	#sum((s2-s3)^2) == 38
	15	s = list(s1, s2, s3)
	16	series = matrix(nrow=L, ncol=n)
	17	for (i in seq_len(n))
	18	series[,i] = s[[I(i,K)]] + rnorm(L,sd=0.01)
	19	getRefSeries = function(indices) {
	20	indices = indices[indices <= n]
	21	if (length(indices)>0) series[,indices] else NULL
	22	}
	23	synchrones = computeSynchrones(bigmemory::as.big.matrix(cbind(s1,s2,s3)), getRefSeries,
	24	n, 100, sync_mean=TRUE, verbose=TRUE, parll=FALSE)
	25
	26	expect_equal(dim(synchrones), c(L,K))
	27	for (i in 1:K)
	28	expect_equal(synchrones[,i], s[[i]], tolerance=0.01)
	29	})
	30
	31	# Helper function to divide indices into balanced sets
	32	test_that("Helper function to spread indices work properly",
	33	{
	34	indices <- 1:400
	35
	36	# bigger nb_per_set than length(indices)
	37	expect_equal(epclust:::.spreadIndices(indices,500), list(indices))
	38
	39	# nb_per_set == length(indices)
	40	expect_equal(epclust:::.spreadIndices(indices,400), list(indices))
	41
	42	# length(indices) %% nb_per_set == 0
	43	expect_equal(epclust:::.spreadIndices(indices,200),
	44	c( list(indices[1:200]), list(indices[201:400]) ))
	45	expect_equal(epclust:::.spreadIndices(indices,100),
	46	c( list(indices[1:100]), list(indices[101:200]),
	47	list(indices[201:300]), list(indices[301:400]) ))
	48
	49	# length(indices) / nb_per_set == 1, length(indices) %% nb_per_set == 100
	50	expect_equal(epclust:::.spreadIndices(indices,300), list(indices))
	51	# length(indices) / nb_per_set == 2, length(indices) %% nb_per_set == 42
	52	repartition <- epclust:::.spreadIndices(indices,179)
	53	expect_equal(length(repartition), 2)
	54	expect_equal(length(repartition[[1]]), 179 + 21)
	55	expect_equal(length(repartition[[1]]), 179 + 21)
	56	})
	57
	58	test_that("clusteringTask1 behave as expected",
	59	{
	60	n = 900
	61	x = seq(0,9.5,0.1)
	62	L = length(x) #96 1/4h
	63	K1 = 60
	64	s = lapply( seq_len(K1), function(i) x^(1+i/30)*cos(x+i) )
	65	series = matrix(nrow=L, ncol=n)
	66	for (i in seq_len(n))
	67	series[,i] = s[[I(i,K1)]] + rnorm(L,sd=0.01)
	68	getSeries = function(indices) {
	69	indices = indices[indices <= n]
	70	if (length(indices)>0) series[,indices] else NULL
	71	}
	72	wf = "haar"
	73	ctype = "absolute"
	74	getContribs = function(indices) curvesToContribs(series[,indices],wf,ctype)
	75	require("cluster", quietly=TRUE)
	76	browser()
	77	algoClust1 = function(contribs,K) cluster::pam(contribs,K,diss=FALSE)$id.med
	78	indices1 = clusteringTask1(1:n, getContribs, K1, algoClust1, 75, verbose=TRUE, parll=FALSE)
	79	medoids_K1 = getSeries(indices1)
	80
	81	expect_equal(dim(medoids_K1), c(L,K1))
	82	# Not easy to evaluate result: at least we expect it to be better than random selection of
	83	# medoids within initial series
	84	distorGood = computeDistortion(series, medoids_K1)
	85	for (i in 1:3)
	86	expect_lte( distorGood, computeDistortion(series,series[,sample(1:n, K1)]) )
	87	})
	88
	89	test_that("clusteringTask2 behave as expected",
	90	{
	91	n = 900
	92	x = seq(0,9.5,0.1)
	93	L = length(x) #96 1/4h
	94	K1 = 60
	95	K2 = 3
	96	#for (i in 1:60) {plot(x^(1+i/30)*cos(x+i),type="l",col=i,ylim=c(-50,50)); par(new=TRUE)}
	97	s = lapply( seq_len(K1), function(i) x^(1+i/30)*cos(x+i) )
	98	series = matrix(nrow=L, ncol=n)
	99	for (i in seq_len(n))
	100	series[i,] = s[[I(i,K1)]] + rnorm(L,sd=0.01)
	101	getRefSeries = function(indices) {
	102	indices = indices[indices <= n]
	103	if (length(indices)>0) series[,indices] else NULL
	104	}
	105	# Artificially simulate 60 medoids - perfect situation, all equal to one of the refs
	106	medoids_K1 = bigmemory::as.big.matrix( sapply( 1:K1, function(i) s[[I(i,K1)]] ) )
	107	medoids_K2 = clusteringTask2(medoids_K1, K2, getRefSeries, n, 75, verbose=TRUE, parll=FALSE)
	108
	109	expect_equal(dim(medoids_K2), c(L,K2))
	110	# Not easy to evaluate result: at least we expect it to be better than random selection of
	111	# medoids within 1...K1 (among references)
	112	distorGood = computeDistortion(series, medoids_K2)
	113	for (i in 1:3)
	114	expect_lte( distorGood, computeDistortion(series,medoids_K1[,sample(1:K1, K2)]) )
	115	})