tests/testthat/acceptance/gak.R

In asardaes/dtwclust: Time Series Clustering Along with Optimizations for the Dynamic Time Warping Distance

context("    Symmetric GAK")

# ==================================================================================================
# setup
# ==================================================================================================

## Original objects in env
ols <- ls()

# ==================================================================================================
# symmetric gak
# ==================================================================================================

test_that("Symmetric univariate GAK distance gives expected results.", {
    D1 <- proxy::dist(data_subset, method = "gak", window.size = 18L)
    sigma <- attr(D1, "sigma")
    D2 <- proxy::dist(data_subset, data_subset, method = "gak", window.size = 18L, sigma = sigma)
    D3 <- sapply(data_subset, GAK, y = data_subset[[1L]], window.size = 18L, sigma = sigma)

    expect_equal(D1, D2, info = "single-double-arg", check.attributes = FALSE)
    expect_equal(D1[, 1L], D3, info = "manual-vs-proxy", check.attributes = FALSE)
})

test_that("Symmetric multivariate GAK distance gives expected results.", {
    D1 <- proxy::dist(data_multivariate, method = "gak", window.size = 18L)
    sigma <- attr(D1, "sigma")
    D2 <- proxy::dist(data_multivariate, data_multivariate, method = "gak", window.size = 18L, sigma = sigma)
    D3 <- sapply(data_multivariate, GAK, y = data_multivariate[[1L]], window.size = 18L, sigma = sigma)

    expect_equal(D1, D2, info = "single-double-arg", check.attributes = FALSE)
    expect_equal(D1[, 1L], D3, info = "manual-vs-proxy", check.attributes = FALSE)
})

# ==================================================================================================
# clean
# ==================================================================================================

rm(list = setdiff(ls(), ols))