epclust/tests/testthat/test-clustering.R

   1 context("clustering")
   2
   3 test_that("clusteringTask1 behave as expected",
   4 {
   5     # Generate 60 reference sinusoïdal series (medoids to be found),
   6     # and sample 900 series around them (add a small noise)
   7     n <- 900
   8     x <- seq(0,9.5,0.1)
   9     L <- length(x) #96 1/4h
  10     K1 <- 60
  11     s <- lapply( seq_len(K1), function(i) x^(1+i/30)*cos(x+i) )
  12     series <- matrix(nrow=L, ncol=n)
  13     for (i in seq_len(n))
  14         series[,i] <- s[[I(i,K1)]] + rnorm(L,sd=0.01)
  15
  16     getSeries <- function(indices) {
  17         indices <- indices[indices <= n]
  18         if (length(indices)>0) as.matrix(series[,indices]) else NULL
  19     }
  20
  21     wf <- "haar"
  22     ctype <- "absolute"
  23     getContribs <- function(indices) curvesToContribs(as.matrix(series[,indices]),wf,ctype)
  24
  25     require("cluster", quietly=TRUE)
  26     algoClust1 <- function(contribs,K) cluster::pam(t(contribs),K,diss=FALSE)$id.med
  27     indices1 <- clusteringTask1(1:n, getContribs, K1, algoClust1, 140, verbose=TRUE)
  28     medoids_K1 <- getSeries(indices1)
  29
  30     expect_equal(dim(medoids_K1), c(L,K1))
  31     # Not easy to evaluate result: at least we expect it to be better than random selection of
  32     # medoids within initial series
  33     distor_good <- computeDistortion(series, medoids_K1)
  34     for (i in 1:3)
  35         expect_lte( distor_good, computeDistortion(series,series[,sample(1:n, K1)]) )
  36 })
  37
  38 test_that("clusteringTask2 behave as expected",
  39 {
  40     # Same 60 reference sinusoïdal series than in clusteringTask1 test,
  41     # but this time we consider them as medoids - skipping stage 1
  42     # Here also we sample 900 series around the 60 "medoids"
  43     n <- 900
  44     x <- seq(0,9.5,0.1)
  45     L <- length(x) #96 1/4h
  46     K1 <- 60
  47     K2 <- 3
  48     #for (i in 1:60) {plot(x^(1+i/30)*cos(x+i),type="l",col=i,ylim=c(-50,50)); par(new=TRUE)}
  49     s <- lapply( seq_len(K1), function(i) x^(1+i/30)*cos(x+i) )
  50     series <- matrix(nrow=L, ncol=n)
  51     for (i in seq_len(n))
  52         series[,i] <- s[[I(i,K1)]] + rnorm(L,sd=0.01)
  53
  54     getSeries <- function(indices) {
  55         indices <- indices[indices <= n]
  56         if (length(indices)>0) as.matrix(series[,indices]) else NULL
  57     }
  58
  59     # Perfect situation: all medoids "after stage 1" are ~good
  60     algoClust2 <- function(dists,K) cluster::pam(dists,K,diss=TRUE)$id.med
  61     indices2 <- clusteringTask2(1:K1, getSeries, K2, algoClust2, 210, 3, 4, 8, "little",
  62         verbose=TRUE)
  63     medoids_K2 <- getSeries(indices2)
  64
  65     expect_equal(dim(medoids_K2), c(L,K2))
  66     # Not easy to evaluate result: at least we expect it to be better than random selection of
  67     # synchrones within 1...K1 (from where distances computations + clustering was run)
  68     distor_good <- computeDistortion(series, medoids_K2)
  69 #TODO: This fails; why?
  70 #   for (i in 1:3)
  71 #       expect_lte( distor_good, computeDistortion(series, series[,sample(1:K1,3)]) )
  72 })