context("collap") if(!is.null(attributes(identical(FALSE, TRUE)))) stop("OECD label issue") bsum <- base::sum bmean <- base::mean # rm(list = ls()) options(warn = -1) g <- GRP(wlddev, ~ country + decade) oa <- function(x) setAttrib(unattrib(x), attributes(x)[c("names", "row.names", "class")]) # Should use above, but sometimes still gives errors if(Sys.getenv("NCRAN") != "TRUE") oa <- function(x) setNames(unattrib(x), names(x)) Mode <- function(x, na.rm = FALSE) { if(na.rm) x <- x[!is.na(x)] ux <- unique(x) ux[which.max(tabulate(match(x, ux)))] } # TODO: What about other return options and weighted multi-function aggregation ? And what about grouped_df method.. test_that("collap performs as intended in simple uses", { expect_equal(collap(mtcars, mtcars$cyl, keep.by = FALSE), fmean(mtcars, mtcars$cyl, use.g.names = FALSE)) expect_equal(collap(mtcars, mtcars[2], keep.by = FALSE), fmean(mtcars, mtcars$cyl, use.g.names = FALSE)) expect_equal(collap(mtcars, ~cyl), fmean(mtcars, mtcars$cyl, use.g.names = FALSE)) expect_equal(collap(mtcars, ~cyl, keep.by = FALSE), fmean(mtcars[-2], mtcars$cyl, use.g.names = FALSE)) expect_equal(collap(iris, ~Species, keep.by = FALSE), fmean(iris[-5], iris$Species, use.g.names = FALSE)) expect_equal(collap(airquality, ~Month, keep.by = FALSE), fmean(airquality[-5], airquality$Month, use.g.names = FALSE)) expect_equal(oa(collap(wlddev, ~ country + decade, keep.col.order = FALSE)), oa(cbind(g$groups, fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))) expect_equal(oa(collap(wlddev, ~ country + decade)), oa(cbind(g$groups, fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))[order(c(1,5,4,9:13,2:3,6:8))]) expect_equal(oa(collap(wlddev, ~ country + decade, keep.col.order = FALSE, keep.by = FALSE)), oa(cbind(fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))) expect_equal(oa(collap(wlddev, ~ country + decade, keep.by = FALSE)), oa(cbind(fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))[order(c(4,9:13,2:3,6:8))]) expect_equal(oa(collap(wlddev, g, keep.by = FALSE)), oa(cbind(g$groups, fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))[order(c(1,5,4,9:13,2:3,6:8))]) expect_equal(oa(collap(wlddev, wlddev[c("country","decade")], keep.by = FALSE)), oa(cbind(g$groups, fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))[order(c(1,5,4,9:13,2:3,6:8))]) }) test_that("collap preserves data attributes", { expect_identical(lapply(collap(wlddev, ~country), attributes), lapply(wlddev, attributes)) expect_identical(vclasses(collap(wlddev, ~country, fmin)), vclasses(wlddev)) expect_identical(vtypes(collap(wlddev, ~country, fmax)), vtypes(wlddev)) expect_identical(lapply(collap(wlddev, ~iso3c), attributes), lapply(wlddev, attributes)) expect_identical(vclasses(collap(wlddev, ~iso3c, fmin)), vclasses(wlddev)) expect_identical(vtypes(collap(wlddev, ~iso3c, fmax)), vtypes(wlddev)) expect_identical(lapply(collap(wlddev, ~date), attributes), lapply(wlddev, attributes)) expect_identical(vclasses(collap(wlddev, ~date, fmin)), vclasses(wlddev)) expect_identical(vtypes(collap(wlddev, ~date, fmax)), vtypes(wlddev)) expect_identical(lapply(collap(wlddev, ~country + decade), attributes), lapply(wlddev, attributes)) expect_identical(vclasses(collap(wlddev, ~country + decade, fmin)), vclasses(wlddev)) expect_identical(vtypes(collap(wlddev, ~country + decade, fmax)), vtypes(wlddev)) }) # if(Sys.getenv("NCRAN") == "TRUE") test_that("collap performs as intended in simple uses with base/stats functions", { expect_equal(oa(collap(mtcars, mtcars$cyl, bsum, keep.by = FALSE)), oa(fsum(mtcars, mtcars$cyl, use.g.names = FALSE))) expect_equal(oa(collap(mtcars, ~cyl, mean.default)), oa(fmean(mtcars, mtcars$cyl, use.g.names = FALSE))) expect_equal(oa(collap(mtcars, ~cyl, bmean)), oa(fmean(mtcars, mtcars$cyl, use.g.names = FALSE))) expect_equal(oa(collap(mtcars, mtcars[2], bsum, keep.by = FALSE)), oa(fsum(mtcars, mtcars$cyl, use.g.names = FALSE))) expect_equal(oa(collap(mtcars, ~cyl, bsum, keep.by = FALSE)), oa(fsum(mtcars[-2], mtcars$cyl, use.g.names = FALSE))) expect_equal(unattrib(collap(iris, ~Species, bsum, keep.by = FALSE)), unattrib(fsum(iris[-5], iris$Species, use.g.names = FALSE))) expect_equal(oa(collap(airquality, ~Month, bsum, na.rm = TRUE, keep.by = FALSE)), oa(fsum(airquality[-5], airquality$Month, use.g.names = FALSE))) expect_equal(oa(collap(wlddev, ~ country + decade, bsum, Mode, na.rm = TRUE, keep.col.order = FALSE)), oa(cbind(g$groups, BY(get_vars(wlddev, c(4,9:13)), g, bsum, na.rm = TRUE, use.g.names = FALSE), BY(get_vars(wlddev, c(2:3,6:8)), g, Mode, na.rm = TRUE, use.g.names = FALSE)))) expect_equal(oa(collap(wlddev, ~ country + decade, bsum, Mode, na.rm = TRUE)), oa(cbind(g$groups, BY(get_vars(wlddev, c(4,9:13)), g, bsum, na.rm = TRUE, use.g.names = FALSE), BY(get_vars(wlddev, c(2:3,6:8)), g, Mode, na.rm = TRUE, use.g.names = FALSE)))[order(c(1,5,4,9:13,2:3,6:8))]) }) test_that("collap using 2-sided formula or cols performs as intended", { expect_equal(oa(collap(mtcars, mpg ~ cyl, keep.by = FALSE)), oa(fmean(mtcars["mpg"], mtcars$cyl, use.g.names = FALSE))) expect_equal(oa(collap(mtcars, mpg ~ cyl, keep.by = FALSE, cols = 300:1000)), oa(fmean(mtcars["mpg"], mtcars$cyl, use.g.names = FALSE))) # cols is ignored, as should be expect_equal(oa(collap(mtcars, ~ cyl, keep.by = FALSE, cols = 1)), oa(fmean(mtcars["mpg"], mtcars$cyl, use.g.names = FALSE))) expect_equal(oa(collap(mtcars, wt + mpg ~ cyl + vs + am, keep.by = FALSE)), oa(fmean(mtcars[c("mpg","wt")], mtcars[c("cyl","vs","am")], use.g.names = FALSE))) expect_equal(oa(collap(mtcars, ~ cyl + vs + am, keep.by = FALSE, cols = c(6,1))), oa(fmean(mtcars[c("mpg","wt")], mtcars[c("cyl","vs","am")], use.g.names = FALSE))) expect_equal(oa(collap(iris, Sepal.Length + Sepal.Width ~ Species, keep.by = FALSE)), oa(fmean(iris[1:2], iris$Species, use.g.names = FALSE))) expect_equal(oa(collap(airquality, ~ Month, keep.by = FALSE)), oa(fmean(airquality[-5], airquality$Month, use.g.names = FALSE))) expect_equal(oa(collap(airquality, ~ Month, keep.by = FALSE, cols = 1:3)), oa(fmean(airquality[1:3], airquality$Month, use.g.names = FALSE))) expect_equal(oa(collap(wlddev, ~ country + decade, cols = 9:13)), oa(collap(wlddev, PCGDP + LIFEEX + GINI + ODA + POP ~ country + decade))) expect_equal(oa(collap(wlddev, ~ country + decade, cols = 9:13)), oa(collap(wlddev, ~ country + decade, cols = 9:13, keep.col.order = FALSE))) expect_equal(oa(collap(wlddev, ~ country + decade, cols = c(2:3,6:8))), oa(collap(wlddev, iso3c + date + region + income + OECD ~ country + decade))) expect_false(identical(oa(collap(wlddev, ~ country + decade, cols = c(2:3,6:8))), oa(collap(wlddev, ~ country + decade, cols = c(2:3,6:8), keep.col.order = FALSE)))) expect_equal(oa(collap(wlddev, ~ country + decade, cols = 9:12, keep.by = FALSE)), oa(collap(wlddev, PCGDP + LIFEEX + GINI + ODA ~ country + decade, keep.by = FALSE))) expect_equal(oa(collap(wlddev, ~ country + decade, cols = 9:13, keep.by = FALSE)), oa(collap(wlddev, ~ country + decade, cols = 9:13, keep.col.order = FALSE, keep.by = FALSE))) expect_equal(oa(collap(wlddev, ~ country + decade, cols = c(2:3,6:8), keep.by = FALSE)), oa(collap(wlddev, iso3c + date + region + income + OECD ~ country + decade, keep.by = FALSE))) expect_equal(oa(collap(wlddev, ~ country + decade, cols = c(2:3,6:8), keep.by = FALSE)), oa(collap(wlddev, ~ country + decade, cols = c(2:3,6:8), keep.col.order = FALSE, keep.by = FALSE))) expect_equal(oa(collap(wlddev, g, cols = 9:12, keep.by = FALSE)), oa(collap(wlddev, PCGDP + LIFEEX + GINI + ODA ~ country + decade, keep.by = FALSE))) expect_equal(oa(collap(wlddev, g, cols = 9:13, keep.by = FALSE)), oa(collap(wlddev, ~ country + decade, cols = 9:13, keep.col.order = FALSE, keep.by = FALSE))) expect_equal(oa(collap(wlddev, g, cols = c(2:3,6:8), keep.by = FALSE)), oa(collap(wlddev, iso3c + date + region + income + OECD ~ country + decade, keep.by = FALSE))) expect_equal(oa(collap(wlddev, g, cols = c(2:3,6:8), keep.by = FALSE)), oa(collap(wlddev, ~ country + decade, cols = c(2:3,6:8), keep.col.order = FALSE, keep.by = FALSE))) expect_equal(oa(collap(wlddev, wlddev[c("country","decade")], cols = 9:12, keep.by = FALSE)), oa(collap(wlddev, PCGDP + LIFEEX + GINI + ODA ~ country + decade, keep.by = FALSE))) expect_equal(oa(collap(wlddev, wlddev[c("country","decade")], cols = 9:13, keep.by = FALSE)), oa(collap(wlddev, ~ country + decade, cols = 9:13, keep.col.order = FALSE, keep.by = FALSE))) expect_equal(oa(collap(wlddev, wlddev[c("country","decade")], cols = c(2:3,6:8), keep.by = FALSE)), oa(collap(wlddev, iso3c + date + region + income + OECD ~ country + decade, keep.by = FALSE))) expect_equal(oa(collap(wlddev, wlddev[c("country","decade")], cols = c(2:3,6:8), keep.by = FALSE)), oa(collap(wlddev, ~ country + decade, cols = c(2:3,6:8), keep.col.order = FALSE, keep.by = FALSE))) }) test_that("collap multi-function aggreagtion performs as intended", { expect_equal(oa(collap(wlddev, ~ country + decade, list(fmean, fmedian), keep.col.order = FALSE, give.names = FALSE)), oa(cbind(g$groups, fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmedian(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))) if(Sys.getenv("NCRAN") == "TRUE") expect_equal(oa(collap(wlddev, ~ country + decade, list(fmean, fmedian), list(fmode, flast), keep.col.order = FALSE, give.names = FALSE)), oa(cbind(g$groups, fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmedian(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE), flast(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))) # with column ordering: expect_equal(unname(oa(collap(wlddev, ~ country + decade, list(fmean, fmedian)))), unname(oa(cbind(g$groups, fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmedian(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE))[order(c(1,5,4,9:13,4,9:13,2:3,6:8))]))) expect_equal(unname(oa(collap(wlddev, ~ country + decade, list(fmean, fmedian), list(fmode, flast)))), unname(oa(cbind(g$groups, fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmedian(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE), flast(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))[order(c(1,5,4,9:13,4,9:13,2:3,6:8,2:3,6:8))])) }) test_that("collap custom aggreagtion performs as intended", { expect_equal(unname(oa(collap(wlddev, ~ country + decade, custom = list(fmean = 9:13, fsd = 9:10, fmode = 7:8), keep.col.order = FALSE))), unname(oa(cbind(g$groups, fmean(wlddev[9:13], g, use.g.names = FALSE), fsd(wlddev[9:10], g, use.g.names = FALSE), fmode(wlddev[7:8], g, use.g.names = FALSE))))) expect_equal(unname(oa(collap(wlddev, ~ country + decade, custom = list(fmean = 9:13, fsd = 9:10, fmode = 7:8)))), unname(oa(cbind(g$groups, fmean(wlddev[9:13], g, use.g.names = FALSE), fsd(wlddev[9:10], g, use.g.names = FALSE), fmode(wlddev[7:8], g, use.g.names = FALSE)))[order(c(1,5,9:13,9:10,7:8))])) expect_equal(oa(collap(wlddev, ~ country + decade, custom = list(fmean = 9:13, fsd = 9:10, fmode = 7:8))), oa(collap(wlddev, ~ country + decade, custom = list(fmean = 9:13, fsd = c("PCGDP","LIFEEX"), fmode = 7:8)))) expect_equal(oa(collap(wlddev, ~ country + decade, custom = list(fmean = "PCGDP", fsd = c("LIFEEX","GINI"), flast = "date"))), oa(collap(wlddev, ~ country + decade, custom = list(fmean = "PCGDP", fsd = 10:11, flast = "date")))) expect_equal(oa(collap(wlddev, g, custom = list(fmean = 9:13, fsd = 9:10, fmode = 7:8))), oa(collap(wlddev, ~ country + decade, custom = list(fmean = 9:13, fsd = c("PCGDP","LIFEEX"), fmode = 7:8)))) expect_equal(oa(collap(wlddev, g, custom = list(fmean = "PCGDP", fsd = c("LIFEEX","GINI"), flast = "date"))), oa(collap(wlddev, g, custom = list(fmean = "PCGDP", fsd = 10:11, flast = "date")))) expect_equal(names(collap(wlddev, g, custom = list(fmean = c(GDP = "PCGDP"), fsd = c("LIFEEX", GN = "GINI"), flast = "date"), keep.by = FALSE, keep.col.order = FALSE)), .c(GDP, LIFEEX, GN, date)) }) test_that("collap weighted aggregations work as intended", { # Not keeping order ... expect_equal(oa(collap(wlddev, ~ country + decade, w = ~ POP, keep.col.order = FALSE)), oa(add_vars(g$groups, fsum(get_vars(wlddev, "POP"), g, use.g.names = FALSE), fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))) expect_equal(oa(collap(wlddev, ~ country + decade, w = ~ POP, keep.col.order = FALSE, keep.by = FALSE)), oa(add_vars(fsum(get_vars(wlddev, "POP"), g, use.g.names = FALSE), fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))) expect_equal(oa(collap(wlddev, ~ country + decade, w = ~ POP, keep.col.order = FALSE, keep.w = FALSE)), oa(add_vars(g$groups, fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))) expect_equal(oa(collap(wlddev, ~ country + decade, w = ~ POP, keep.col.order = FALSE, keep.by = FALSE, keep.w = FALSE)), oa(add_vars(fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))) # keeping order ... expect_equal(oa(collap(wlddev, ~ country + decade, w = ~ POP)), oa(add_vars(g$groups, fsum(get_vars(wlddev, "POP"), g, use.g.names = FALSE), fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))[names(wlddev)]) expect_equal(oa(collap(wlddev, ~ country + decade, w = ~ POP, keep.by = FALSE)), oa(add_vars(fsum(get_vars(wlddev, "POP"), g, use.g.names = FALSE), fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))[setdiff(names(wlddev), g$group.vars)]) expect_equal(oa(collap(wlddev, ~ country + decade, w = ~ POP, keep.w = FALSE)), oa(add_vars(g$groups, fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))[setdiff(names(wlddev), "POP")]) expect_equal(oa(collap(wlddev, ~ country + decade, w = ~ POP, keep.by = FALSE, keep.w = FALSE)), oa(add_vars(fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))[setdiff(names(wlddev), c(g$group.vars, "POP"))]) }) if(Sys.getenv("NCRAN") == "TRUE") test_that("collap multi-function aggreagtion with weights performs as intended", { expect_equal(oa(collap(wlddev, ~ country + decade, list(fmean, fsd), w = ~ POP, keep.col.order = FALSE, give.names = FALSE)), oa(cbind(g$groups, fsum(get_vars(wlddev, 13), g, use.g.names = FALSE), fmean(get_vars(wlddev, c(4,9:12)), g, wlddev$POP, use.g.names = FALSE), fsd(get_vars(wlddev, c(4,9:12)), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, wlddev$POP, use.g.names = FALSE)))) expect_equal(oa(collap(wlddev, ~ country + decade, list(fmean, fsd), list(fmode, flast), w = ~ POP, keep.col.order = FALSE, give.names = FALSE)), oa(cbind(g$groups, fsum(get_vars(wlddev, 13), g, use.g.names = FALSE), fmean(get_vars(wlddev, c(4,9:12)), g, wlddev$POP, use.g.names = FALSE), fsd(get_vars(wlddev, c(4,9:12)), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, wlddev$POP, use.g.names = FALSE), flast(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))) # with column ordering: expect_equal(unname(oa(collap(wlddev, ~ country + decade, list(fmean, fsd), w = ~ POP, wFUN = list(fsum, fmax)))), unname(oa(cbind(g$groups, fsum(get_vars(wlddev, 13), g, use.g.names = FALSE), fmax(get_vars(wlddev, 13), g, use.g.names = FALSE), fmean(get_vars(wlddev, c(4,9:12)), g, wlddev$POP, use.g.names = FALSE), fsd(get_vars(wlddev, c(4,9:12)), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, wlddev$POP, use.g.names = FALSE)))[order(c(1,5,13,13,4,9:12,4,9:12,2:3,6:8))])) expect_equal(unname(oa(collap(wlddev, ~ country + decade, list(fmean, fsd), list(fmode, flast), w = ~ POP, wFUN = list(fsum, fmax)))), unname(oa(cbind(g$groups, fsum(get_vars(wlddev, 13), g, use.g.names = FALSE), fmax(get_vars(wlddev, 13), g, use.g.names = FALSE), fmean(get_vars(wlddev, c(4,9:12)), g, wlddev$POP, use.g.names = FALSE), fsd(get_vars(wlddev, c(4,9:12)), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, wlddev$POP, use.g.names = FALSE), flast(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))[order(c(1,5,13,13,4,9:12,4,9:12,2:3,6:8,2:3,6:8))])) }) v1 <- c("year","PCGDP","LIFEEX","GINI","ODA") v2 <- c("iso3c","date","region","income", "OECD") test_that("collap weighted customized aggregation works as intended", { # Not keeping order ... expect_equal(oa(collap(wlddev, ~ country + decade, w = ~ POP, custom = list(fmean = v1, fmode = v2), keep.col.order = FALSE, give.names = FALSE)), oa(add_vars(g$groups, fsum(get_vars(wlddev, "POP"), g, use.g.names = FALSE), fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))) expect_equal(unattrib(collap(wlddev, ~ country + decade, w = ~ POP, custom = list(fmean = v1, fmode = v2), keep.col.order = FALSE, keep.by = FALSE, give.names = FALSE)), unattrib(add_vars(fsum(get_vars(wlddev, "POP"), g, use.g.names = FALSE), fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))) expect_equal(oa(collap(wlddev, ~ country + decade, w = ~ POP, custom = list(fmean = v1, fmode = v2), keep.col.order = FALSE, keep.w = FALSE, give.names = FALSE)), oa(add_vars(g$groups, fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))) expect_equal(oa(collap(wlddev, ~ country + decade, w = ~ POP, custom = list(fmean = v1, fmode = v2), keep.col.order = FALSE, keep.by = FALSE, keep.w = FALSE, give.names = FALSE)), oa(add_vars(fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))) # keeping order ... expect_equal(oa(collap(wlddev, ~ country + decade, w = ~ POP, custom = list(fmean = v1, fmode = v2), give.names = FALSE)), oa(add_vars(g$groups, fsum(get_vars(wlddev, "POP"), g, use.g.names = FALSE), fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))[names(wlddev)]) expect_equal(oa(collap(wlddev, ~ country + decade, w = ~ POP, keep.by = FALSE, custom = list(fmean = v1, fmode = v2), give.names = FALSE)), oa(add_vars(fsum(get_vars(wlddev, "POP"), g, use.g.names = FALSE), fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))[setdiff(names(wlddev), g$group.vars)]) expect_equal(oa(collap(wlddev, ~ country + decade, w = ~ POP, keep.w = FALSE, custom = list(fmean = v1, fmode = v2), give.names = FALSE)), oa(add_vars(g$groups, fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))[setdiff(names(wlddev), "POP")]) expect_equal(oa(collap(wlddev, ~ country + decade, w = ~ POP, keep.by = FALSE, keep.w = FALSE, custom = list(fmean = v1, fmode = v2), give.names = FALSE)), oa(add_vars(fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))[setdiff(names(wlddev), c(g$group.vars, "POP"))]) }) test_that("collap gives informative errors", { expect_error(collap(~cyl, ~cyl)) # nah, need to give error in qDF expect_error(collap(wlddev, 1:3)) # only gives error in fmean.. a bit late.. expect_error(collap(wlddev, "country")) # same thing expect_error(collap(wlddev, ~ country1)) expect_error(collap(wlddev, ~ country, w = ~bla)) expect_error(collap(wlddev, ~ country, w = ~POP, wFUN = bsum)) expect_error(collap(wlddev, ~ country + year + bla)) expect_error(collap(wlddev, bla ~ country)) expect_warning(collap(wlddev, ~ country, bla = 1)) # passes to fmean.data.frame which give the error. # expect_error(collap(wlddev, ~ country, bsum, cols = 9:13, bla = 1)) # This is an issue, bsum(1:3, bla = 1) does not give an error expect_error(collap(wlddev, mtcars$cyl)) # again fmean error.. expect_error(collap(wlddev, ~iso3c, cols = 9:14)) # expect_error(collap(wlddev, ~iso3c, cols = 0:1)) # no error.. expect_error(collap(wlddev, ~iso3c, cols = c("PCGDP","bla"))) expect_error(collap(wlddev, ~iso3c, cols = c("PCGDP","LIFEEX1"))) expect_error(collap(wlddev, ~iso3c, custom = ~ PCGDP)) expect_error(collap(wlddev, ~iso3c, custom = list(fmean, fmode))) expect_error(collap(wlddev, ~iso3c, custom = list(fmean = 9:14, fmode = 4:6))) expect_error(collap(wlddev, ~iso3c, custom = list(fmean = 9:13, 4:6))) expect_error(collap(wlddev, ~iso3c, custom = list(fmean = 9:13, fmode2 = 4:6))) expect_error(collap(wlddev, ~iso3c, custom = list(fmean = 9:13, fmode = c("GINI","bla")))) expect_error(collap(wlddev, ~iso3c, custom = list(fmean = 9:13, fmode = c("GINI","PCGDP2")))) }) # Note: one more thing to test is performance with vector-valued functions... # Testing collapv v <- c(1, 5) test_that("collapv performs as intended in simple uses", { expect_equal(oa(collapv(mtcars, 2)), oa(fmean(mtcars, mtcars$cyl, use.g.names = FALSE))) expect_equal(oa(collapv(mtcars, 2, keep.by = FALSE)), oa(fmean(mtcars[-2], mtcars$cyl, use.g.names = FALSE))) expect_equal(oa(collapv(iris, "Species", keep.by = FALSE)), oa(fmean(iris[-5], iris$Species, use.g.names = FALSE))) expect_equal(oa(collapv(airquality, "Month", keep.by = FALSE)), oa(fmean(airquality[-5], airquality$Month, use.g.names = FALSE))) expect_equal(oa(collapv(wlddev, v, keep.col.order = FALSE)), oa(cbind(g$groups, fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))) expect_equal(oa(collapv(wlddev, v)), oa(cbind(g$groups, fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))[order(c(1,5,4,9:13,2:3,6:8))]) expect_equal(oa(collapv(wlddev, v, keep.col.order = FALSE, keep.by = FALSE)), oa(cbind(fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))) expect_equal(oa(collapv(wlddev, v, keep.by = FALSE)), oa(cbind(fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))[order(c(4,9:13,2:3,6:8))]) expect_equal(names(collapv(wlddev, v, custom = list(fmean = c(GDP = "PCGDP"), fsd = c("LIFEEX", GN = "GINI"), flast = "date"), keep.by = FALSE, keep.col.order = FALSE)), .c(GDP, LIFEEX, GN, date)) }) test_that("collapv preserves data attributes", { expect_identical(lapply(collapv(wlddev, 1), attributes), lapply(wlddev, attributes)) expect_identical(vclasses(collapv(wlddev, 1, fmax)), vclasses(wlddev)) expect_identical(vtypes(collapv(wlddev, 1, fmin)), vtypes(wlddev)) expect_identical(lapply(collapv(wlddev, "iso3c"), attributes), lapply(wlddev, attributes)) expect_identical(vclasses(collapv(wlddev, "iso3c", fmax)), vclasses(wlddev)) expect_identical(vtypes(collapv(wlddev, "iso3c", fmin)), vtypes(wlddev)) expect_identical(lapply(collapv(wlddev, "date"), attributes), lapply(wlddev, attributes)) expect_identical(vclasses(collapv(wlddev, "date", ffirst)), vclasses(wlddev)) expect_identical(vtypes(collapv(wlddev, "date", flast)), vtypes(wlddev)) expect_identical(lapply(collapv(wlddev, v), attributes), lapply(wlddev, attributes)) expect_identical(vclasses(collapv(wlddev, v, flast)), vclasses(wlddev)) expect_identical(vtypes(collapv(wlddev, v, ffirst)), vtypes(wlddev)) }) # if(Sys.getenv("NCRAN") == "TRUE") test_that("collapv performs as intended in simple uses with base/stats functions", { expect_equal(oa(collapv(mtcars, "cyl", mean.default)), oa(fmean(mtcars, mtcars$cyl, use.g.names = FALSE))) expect_equal(oa(collapv(mtcars, "cyl", bmean)), oa(fmean(mtcars, mtcars$cyl, use.g.names = FALSE))) expect_equal(oa(collapv(mtcars, 2, bsum, keep.by = FALSE)), oa(fsum(mtcars[-2], mtcars$cyl, use.g.names = FALSE))) expect_equal(oa(collapv(iris, 5, bsum, keep.by = FALSE)), oa(fsum(iris[-5], iris$Species, use.g.names = FALSE))) expect_equal(oa(collapv(airquality, "Month", bsum, na.rm = TRUE, keep.by = FALSE)), oa(fsum(airquality[-5], airquality$Month, use.g.names = FALSE))) expect_equal(oa(collapv(wlddev, v, bsum, Mode, na.rm = TRUE, keep.col.order = FALSE)), oa(cbind(g$groups, BY(get_vars(wlddev, c(4,9:13)), g, bsum, na.rm = TRUE, use.g.names = FALSE), BY(get_vars(wlddev, c(2:3,6:8)), g, Mode, na.rm = TRUE, use.g.names = FALSE)))) expect_equal(oa(collapv(wlddev, v, bsum, Mode, na.rm = TRUE)), oa(cbind(g$groups, BY(get_vars(wlddev, c(4,9:13)), g, bsum, na.rm = TRUE, use.g.names = FALSE), BY(get_vars(wlddev, c(2:3,6:8)), g, Mode, na.rm = TRUE, use.g.names = FALSE)))[order(c(1,5,4,9:13,2:3,6:8))]) }) test_that("collapv using cols performs as intended", { expect_equal(oa(collapv(mtcars, 2, keep.by = FALSE, cols = 1)), oa(fmean(mtcars["mpg"], mtcars$cyl, use.g.names = FALSE))) expect_equal(oa(collapv(mtcars, c("cyl", "vs", "am"), keep.by = FALSE, cols = c(6,1))), oa(fmean(mtcars[c("mpg","wt")], mtcars[c("cyl","vs","am")], use.g.names = FALSE))) expect_equal(oa(collapv(airquality, "Month", keep.by = FALSE)), oa(fmean(airquality[-5], airquality$Month, use.g.names = FALSE))) expect_equal(oa(collapv(airquality, "Month", keep.by = FALSE, cols = 1:3)), oa(fmean(airquality[1:3], airquality$Month, use.g.names = FALSE))) expect_equal(oa(collapv(wlddev, v, cols = 9:12)), oa(collap(wlddev, PCGDP + LIFEEX + GINI + ODA ~ country + decade))) expect_equal(oa(collapv(wlddev, v, cols = 9:13)), oa(collap(wlddev, ~ country + decade, cols = 9:13, keep.col.order = FALSE))) expect_equal(oa(collapv(wlddev, v, cols = c(2:3,6:8))), oa(collap(wlddev, iso3c + date + region + income + OECD ~ country + decade))) expect_false(identical(collapv(wlddev, v, cols = c(2:3,6:8)), collap(wlddev, ~ country + decade, cols = c(2:3,6:8), keep.col.order = FALSE))) expect_equal(oa(collapv(wlddev, v, cols = 9:12, keep.by = FALSE)), oa(collap(wlddev, PCGDP + LIFEEX + GINI + ODA ~ country + decade, keep.by = FALSE))) expect_equal(oa(collapv(wlddev, v, cols = 9:13, keep.by = FALSE)), oa(collap(wlddev, ~ country + decade, cols = 9:13, keep.col.order = FALSE, keep.by = FALSE))) expect_equal(oa(collapv(wlddev, v, cols = c(2:3,6:8), keep.by = FALSE)), oa(collap(wlddev, iso3c + date + region + income + OECD ~ country + decade, keep.by = FALSE))) expect_equal(oa(collapv(wlddev, v, cols = c(2:3,6:8), keep.by = FALSE)), oa(collap(wlddev, ~ country + decade, cols = c(2:3,6:8), keep.col.order = FALSE, keep.by = FALSE))) }) test_that("collapv multi-function aggreagtion performs as intended", { expect_equal(oa(collapv(wlddev, v, list(fmean, fmedian), keep.col.order = FALSE, give.names = FALSE)), oa(cbind(g$groups, fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmedian(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))) expect_equal(oa(collapv(wlddev, v, list(fmean, fmedian), list(fmode, flast), keep.col.order = FALSE, give.names = FALSE)), oa(cbind(g$groups, fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmedian(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE), flast(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))) # with column ordering: expect_equal(unname(oa(collapv(wlddev, v, list(fmean, fmedian)))), unname(oa(cbind(g$groups, fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmedian(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))[order(c(1,5,4,9:13,4,9:13,2:3,6:8))])) expect_equal(unname(oa(collapv(wlddev, v, list(fmean, fmedian), list(fmode, flast)))), unname(oa(cbind(g$groups, fmean(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmedian(get_vars(wlddev, c(4,9:13)), g, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE), flast(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))[order(c(1,5,4,9:13,4,9:13,2:3,6:8,2:3,6:8))])) }) test_that("collapv custom aggreagtion performs as intended", { expect_equal(unname(oa(collapv(wlddev, v, custom = list(fmean = 9:13, fsd = 9:10, fmode = 7:8), keep.col.order = FALSE))), unname(oa(cbind(g$groups, fmean(wlddev[9:13], g, use.g.names = FALSE), fsd(wlddev[9:10], g, use.g.names = FALSE), fmode(wlddev[7:8], g, use.g.names = FALSE))))) expect_equal(unname(oa(collapv(wlddev, v, custom = list(fmean = 9:13, fsd = 9:10, fmode = 7:8)))), unname(oa(cbind(g$groups, fmean(wlddev[9:13], g, use.g.names = FALSE), fsd(wlddev[9:10], g, use.g.names = FALSE), fmode(wlddev[7:8], g, use.g.names = FALSE)))[order(c(1,5,9:13,9:10,7:8))])) expect_equal(oa(collapv(wlddev, v, custom = list(fmean = 9:13, fsd = 9:10, fmode = 7:8))), oa(collapv(wlddev, v, custom = list(fmean = 9:13, fsd = c("PCGDP","LIFEEX"), fmode = 7:8)))) expect_equal(oa(collapv(wlddev, v, custom = list(fmean = "PCGDP", fsd = c("LIFEEX","GINI"), flast = "date"))), oa(collapv(wlddev, v, custom = list(fmean = "PCGDP", fsd = 10:11, flast = "date")))) expect_equal(oa(collapv(wlddev, v, custom = list(fmean = 9:13, fsd = 9:10, fmode = 7:8))), oa(collapv(wlddev, v, custom = list(fmean = 9:13, fsd = c("PCGDP","LIFEEX"), fmode = 7:8)))) expect_equal(oa(collapv(wlddev, v, custom = list(fmean = "PCGDP", fsd = c("LIFEEX","GINI"), flast = "date"))), oa(collapv(wlddev, v, custom = list(fmean = "PCGDP", fsd = 10:11, flast = "date")))) }) test_that("collapv weighted aggregations work as intended", { # Not keeping order ... expect_equal(oa(collapv(wlddev, v, w = "POP", keep.col.order = FALSE)), oa(add_vars(g$groups, fsum(get_vars(wlddev, "POP"), g, use.g.names = FALSE), fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))) expect_equal(oa(collapv(wlddev, v, w = "POP", keep.col.order = FALSE, keep.by = FALSE)), oa(add_vars(fsum(get_vars(wlddev, "POP"), g, use.g.names = FALSE), fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))) expect_equal(oa(collapv(wlddev, v, w = "POP", keep.col.order = FALSE, keep.w = FALSE)), oa(add_vars(g$groups, fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))) expect_equal(oa(collapv(wlddev, v, w = "POP", keep.col.order = FALSE, keep.by = FALSE, keep.w = FALSE)), oa(add_vars(fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))) # keeping order ... expect_equal(oa(collapv(wlddev, v, w = "POP")), oa(add_vars(g$groups, fsum(get_vars(wlddev, "POP"), g, use.g.names = FALSE), fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))[names(wlddev)]) expect_equal(unattrib(collapv(wlddev, v, w = "POP", keep.by = FALSE)), unattrib(add_vars(fsum(get_vars(wlddev, "POP"), g, use.g.names = FALSE), fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE))[setdiff(names(wlddev), g$group.vars)])) expect_equal(oa(collapv(wlddev, v, w = "POP", keep.w = FALSE)), oa(add_vars(g$groups, fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))[setdiff(names(wlddev), "POP")]) expect_equal(oa(collapv(wlddev, v, w = "POP", keep.by = FALSE, keep.w = FALSE)), oa(add_vars(fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))[setdiff(names(wlddev), c(g$group.vars, "POP"))]) }) if(Sys.getenv("NCRAN") == "TRUE") test_that("collapv multi-function aggreagtion with weights performs as intended", { expect_equal(oa(collapv(wlddev, v, list(fmean, fsd), w = "POP", keep.col.order = FALSE, give.names = FALSE)), oa(cbind(g$groups, fsum(get_vars(wlddev, 13), g, use.g.names = FALSE), fmean(get_vars(wlddev, c(4,9:12)), g, wlddev$POP, use.g.names = FALSE), fsd(get_vars(wlddev, c(4,9:12)), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, wlddev$POP, use.g.names = FALSE)))) expect_equal(oa(collapv(wlddev, v, list(fmean, fsd), list(fmode, flast), w = "POP", keep.col.order = FALSE, give.names = FALSE)), oa(cbind(g$groups, fsum(get_vars(wlddev, 13), g, use.g.names = FALSE), fmean(get_vars(wlddev, c(4,9:12)), g, wlddev$POP, use.g.names = FALSE), fsd(get_vars(wlddev, c(4,9:12)), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, wlddev$POP, use.g.names = FALSE), flast(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))) # with column ordering: expect_equal(unname(oa(collapv(wlddev, v, list(fmean, fsd), w = "POP", wFUN = list(fsum, fmax)))), unname(oa(cbind(g$groups, fsum(get_vars(wlddev, 13), g, use.g.names = FALSE), fmax(get_vars(wlddev, 13), g, use.g.names = FALSE), fmean(get_vars(wlddev, c(4,9:12)), g, wlddev$POP, use.g.names = FALSE), fsd(get_vars(wlddev, c(4,9:12)), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, wlddev$POP, use.g.names = FALSE)))[order(c(1,5,13,13,4,9:12,4,9:12,2:3,6:8))])) expect_equal(unattrib(collapv(wlddev, v, list(fmean, fsd), list(fmode, flast), w = "POP", wFUN = list(fsum, fmax))), unattrib(cbind(g$groups, fsum(get_vars(wlddev, 13), g, use.g.names = FALSE), fmax(get_vars(wlddev, 13), g, use.g.names = FALSE), fmean(get_vars(wlddev, c(4,9:12)), g, wlddev$POP, use.g.names = FALSE), fsd(get_vars(wlddev, c(4,9:12)), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c(2:3,6:8)), g, wlddev$POP, use.g.names = FALSE), flast(get_vars(wlddev, c(2:3,6:8)), g, use.g.names = FALSE)))[order(c(1,5,13,13,4,9:12,4,9:12,2:3,6:8,2:3,6:8))]) }) v1 <- c("year","PCGDP","LIFEEX","GINI","ODA") v2 <- c("iso3c","date","region","income", "OECD") test_that("collapv weighted customized aggregation works as intended", { # Not keeping order ... expect_equal(oa(collapv(wlddev, v, w = "POP", custom = list(fmean = v1, fmode = v2), keep.col.order = FALSE, give.names = FALSE)), oa(add_vars(g$groups, fsum(get_vars(wlddev, "POP"), g, use.g.names = FALSE), fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))) expect_equal(unattrib(collapv(wlddev, v, w = "POP", custom = list(fmean = v1, fmode = v2), keep.col.order = FALSE, keep.by = FALSE, give.names = FALSE)), unattrib(add_vars(fsum(get_vars(wlddev, "POP"), g, use.g.names = FALSE), fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))) expect_equal(oa(collapv(wlddev, v, w = "POP", custom = list(fmean = v1, fmode = v2), keep.col.order = FALSE, keep.w = FALSE, give.names = FALSE)), oa(add_vars(g$groups, fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))) expect_equal(oa(collapv(wlddev, v, w = "POP", custom = list(fmean = v1, fmode = v2), keep.col.order = FALSE, keep.by = FALSE, keep.w = FALSE, give.names = FALSE)), oa(add_vars(fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))) # keeping order ... expect_equal(oa(collapv(wlddev, v, w = "POP", custom = list(fmean = v1, fmode = v2), give.names = FALSE)), oa(add_vars(g$groups, fsum(get_vars(wlddev, "POP"), g, use.g.names = FALSE), fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))[names(wlddev)]) expect_equal(oa(collapv(wlddev, v, w = "POP", keep.by = FALSE, custom = list(fmean = v1, fmode = v2), give.names = FALSE)), oa(add_vars(fsum(get_vars(wlddev, "POP"), g, use.g.names = FALSE), fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))[setdiff(names(wlddev), g$group.vars)]) expect_equal(oa(collapv(wlddev, v, w = "POP", keep.w = FALSE, custom = list(fmean = v1, fmode = v2), give.names = FALSE)), oa(add_vars(g$groups, fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))[setdiff(names(wlddev), "POP")]) expect_equal(oa(collapv(wlddev, v, w = "POP", keep.by = FALSE, keep.w = FALSE, custom = list(fmean = v1, fmode = v2), give.names = FALSE)), oa(add_vars(fmean(get_vars(wlddev, c("year","PCGDP","LIFEEX","GINI","ODA")), g, wlddev$POP, use.g.names = FALSE), fmode(get_vars(wlddev, c("iso3c","date","region","income", "OECD")), g, wlddev$POP, use.g.names = FALSE)))[setdiff(names(wlddev), c(g$group.vars, "POP"))]) }) test_that("collapv gives informative errors", { expect_error(collapv(~cyl, ~cyl)) # nah, need to give error in qDF expect_error(collapv(wlddev, ~ country)) # same thing expect_error(collapv(wlddev, 14)) expect_error(collapv(wlddev, 1, w = 14)) expect_error(collapv(wlddev, 1, w = "bla")) expect_error(collapv(wlddev, 1, w = 13, wFUN = bsum)) expect_error(collapv(wlddev, c(1,0))) expect_error(collapv(wlddev, c(1,14))) expect_warning(collapv(wlddev, 1, bla = 1)) # passes to fmean.data.frame which give the error. expect_error(collapv(wlddev, 2, cols = 9:14)) expect_error(collapv(wlddev, 2, cols = c("PCGDP","bla"))) expect_error(collapv(wlddev, 2, cols = c("PCGDP","LIFEEX1"))) expect_error(collapv(wlddev, 2, custom = ~ PCGDP)) expect_error(collapv(wlddev, 2, custom = list(fmean, fmode))) expect_error(collapv(wlddev, 2, custom = list(fmean = 9:14, fmode = 4:6))) expect_error(collapv(wlddev, 2, custom = list(fmean = 9:14, 4:6))) expect_error(collapv(wlddev, 2, custom = list(fmean = 9:14, fmode2 = 4:6))) expect_error(collapv(wlddev, 2, custom = list(fmean = 9:13, fmode = c("GINI","bla")))) expect_error(collapv(wlddev, 2, custom = list(fmean = 9:13, fmode = c("GINI","PCGDP2")))) }) options(warn = 1)