context("EOF") data(geopotential) test_that("EOF runs", { expect_s3_class({ EOF(gh ~ lat + lon | date, data = geopotential, n = 1) }, "eof") }) test_that("EOF, returns correct PCs", { expect_equal(nrow(EOF(gh ~ lat + lon | date, data = geopotential, n = 2)$sdev), 1) }) test_that("EOF uses bootstrap", { expect_s3_class({ data(geopotential) EOF(gh ~ lat + lon | date, data = geopotential, n = 1, B = 10) }, "eof") }) test_that("can use differnet engine", { set.seed(40) # with this seed, the base::svd result has a different sign. expect_equal( EOF(gh ~ lat + lon | date, data = geopotential, engine = base::svd)$left[, gh := -gh], EOF(gh ~ lat + lon | date, data = geopotential, n = 1)$left ) }) test_that("EOF works inside data.table", { expect_equal(geopotential[, EOF(gh ~ lon + lat | date)$left], EOF(gh ~ lon + lat | date, data = geopotential)$left) }) test_that("EOF rotates", { expect_identical( round(EOF(gh ~ lon + lat | date, data = geopotential, n = 1:2, rotate = TRUE)$sdev$sd), c(1424982, 542271) ) }) test_that("EOF fails gracefully", { expect_error(EOF(gh ~ lon + lat | date, data = geopotential, fill = "a")) expect_error(EOF(gh ~ lon1 + lat2 | date, data = geopotential), "Columns not found in data: lon1, lat2") expect_error(EOF(gh ~ lon | date, data = geopotential), "The formula gh ~ lon | date does not identify an unique observation for each cell.") }) test_that("eof methods", { eof <- EOF(gh ~ lat + lon | date, data = geopotential, n = 1:5, engine = base::svd) # need to force this engine so that predict is exact eof_12 <- cut(eof, 1:2) expect_s3_class(eof_12, "eof") expect_equal(as.character(unique(eof_12$left$PC)), c("PC1", "PC2")) expect_equal(as.character(unique(eof_12$right$PC)), c("PC1", "PC2")) expect_equal(as.character(unique(eof_12$sdev$PC)), c("PC1", "PC2")) expect_true(inherits(screeplot(eof), "gg")) expect_true(inherits(autoplot(eof), "gg")) eof_all <- EOF(gh ~ lat + lon | date, data = geopotential, n = NULL, engine = base::svd) expect_equal(geopotential[, .(lat, lon, date, gh)][order(lat, lon, date)], predict(eof_all)[order(lat, lon, date)]) expect_equal(predict(eof_all, n = 1:5), predict(eof)) expect_known_output(summary(eof), file = "eof_summary") })