context("miscellaneous issues") # rm(list = ls()) options(warn = -1) F <- getNamespace("collapse")$F if(identical(Sys.getenv("NCRAN"), "TRUE")) { test_that("Using a factor with unused levels does not pose a problem to flag, fdiff or fgrowth (#25)", { wlddev2 <- subset(wlddev, iso3c %in% c("ALB", "AFG", "DZA")) wlddev3 <- droplevels(wlddev2) expect_identical(L(wlddev3, 1, LIFEEX~iso3c, ~year), L(wlddev3, 1, ~iso3c, ~year, cols="LIFEEX")) expect_identical(L(wlddev3, -1:1, LIFEEX~iso3c, ~year), L(wlddev3, -1:1, ~iso3c, ~year, cols="LIFEEX")) expect_identical(droplevels(L(wlddev2, 1, ~iso3c, ~year, cols="LIFEEX")), L(wlddev3, 1, ~iso3c, ~year, cols="LIFEEX")) expect_identical(droplevels(L(wlddev2, -1:1, ~iso3c, ~year, cols="LIFEEX")), L(wlddev3, -1:1, ~iso3c, ~year, cols="LIFEEX")) expect_identical(droplevels(D(wlddev2, 1, 1, ~iso3c, ~year, cols="LIFEEX")), D(wlddev3, 1, 1, ~iso3c, ~year, cols="LIFEEX")) expect_identical(droplevels(D(wlddev2, -1:1, 1:2, ~iso3c, ~year, cols="LIFEEX")), D(wlddev3, -1:1, 1:2, ~iso3c, ~year, cols="LIFEEX")) expect_identical(droplevels(Dlog(wlddev2, 1, 1, ~iso3c, ~year, cols="LIFEEX")), Dlog(wlddev3, 1, 1, ~iso3c, ~year, cols="LIFEEX")) expect_identical(droplevels(Dlog(wlddev2, -1:1, 1:2, ~iso3c, ~year, cols="LIFEEX")), Dlog(wlddev3, -1:1, 1:2, ~iso3c, ~year, cols="LIFEEX")) expect_identical(droplevels(D(wlddev2, 1, 1, ~iso3c, ~year, cols="LIFEEX", rho = 0.95)), D(wlddev3, 1, 1, ~iso3c, ~year, cols="LIFEEX", rho = 0.95)) expect_identical(droplevels(D(wlddev2, -1:1, 1:2, ~iso3c, ~year, cols="LIFEEX", rho = 0.95)), D(wlddev3, -1:1, 1:2, ~iso3c, ~year, cols="LIFEEX", rho = 0.95)) expect_identical(droplevels(Dlog(wlddev2, 1, 1, ~iso3c, ~year, cols="LIFEEX", rho = 0.95)), Dlog(wlddev3, 1, 1, ~iso3c, ~year, cols="LIFEEX", rho = 0.95)) expect_identical(droplevels(Dlog(wlddev2, -1:1, 1:2, ~iso3c, ~year, cols="LIFEEX", rho = 0.95)), Dlog(wlddev3, -1:1, 1:2, ~iso3c, ~year, cols="LIFEEX", rho = 0.95)) expect_identical(droplevels(G(wlddev2, 1, 1, ~iso3c, ~year, cols="LIFEEX")), G(wlddev3, 1, 1, ~iso3c, ~year, cols="LIFEEX")) expect_identical(droplevels(G(wlddev2, -1:1, 1:2, ~iso3c, ~year, cols="LIFEEX")), G(wlddev3, -1:1, 1:2, ~iso3c, ~year, cols="LIFEEX")) expect_identical(L(wlddev3, 1, LIFEEX~iso3c), L(wlddev3, 1, ~iso3c, cols="LIFEEX")) expect_identical(L(wlddev3, -1:1, LIFEEX~iso3c), L(wlddev3, -1:1, ~iso3c, cols="LIFEEX")) expect_identical(droplevels(L(wlddev2, 1, ~iso3c, cols="LIFEEX")), L(wlddev3, 1, ~iso3c, cols="LIFEEX")) expect_identical(droplevels(L(wlddev2, -1:1, ~iso3c, cols="LIFEEX")), L(wlddev3, -1:1, ~iso3c, cols="LIFEEX")) expect_identical(droplevels(D(wlddev2, 1, 1, ~iso3c, cols="LIFEEX")), D(wlddev3, 1, 1, ~iso3c, cols="LIFEEX")) expect_identical(droplevels(D(wlddev2, -1:1, 1:2, ~iso3c, cols="LIFEEX")), D(wlddev3, -1:1, 1:2, ~iso3c, cols="LIFEEX")) expect_identical(droplevels(Dlog(wlddev2, 1, 1, ~iso3c, cols="LIFEEX")), Dlog(wlddev3, 1, 1, ~iso3c, cols="LIFEEX")) expect_identical(droplevels(Dlog(wlddev2, -1:1, 1:2, ~iso3c, cols="LIFEEX")), Dlog(wlddev3, -1:1, 1:2, ~iso3c, cols="LIFEEX")) expect_identical(droplevels(D(wlddev2, 1, 1, ~iso3c, cols="LIFEEX", rho = 0.95)), D(wlddev3, 1, 1, ~iso3c, cols="LIFEEX", rho = 0.95)) expect_identical(droplevels(D(wlddev2, -1:1, 1:2, ~iso3c, cols="LIFEEX", rho = 0.95)), D(wlddev3, -1:1, 1:2, ~iso3c, cols="LIFEEX", rho = 0.95)) expect_identical(droplevels(Dlog(wlddev2, 1, 1, ~iso3c, cols="LIFEEX", rho = 0.95)), Dlog(wlddev3, 1, 1, ~iso3c, cols="LIFEEX", rho = 0.95)) expect_identical(droplevels(Dlog(wlddev2, -1:1, 1:2, ~iso3c, cols="LIFEEX", rho = 0.95)), Dlog(wlddev3, -1:1, 1:2, ~iso3c, cols="LIFEEX", rho = 0.95)) expect_identical(droplevels(G(wlddev2, 1, 1, ~iso3c, cols="LIFEEX")), G(wlddev3, 1, 1, ~iso3c, cols="LIFEEX")) expect_identical(droplevels(G(wlddev2, -1:1, 1:2, ~iso3c, cols="LIFEEX")), G(wlddev3, -1:1, 1:2, ~iso3c, cols="LIFEEX")) }) test_that("Using a factor with unused levels does not pose a problem to statistical functions", { wlddev2 <- fsubset(wlddev, iso3c %in% c("ALB", "AFG", "DZA")) d <- nv(wlddev2) m <- qM(d) v <- d$PCGDP w <- rep(1, length(v)) f <- wlddev2$iso3c lev <- levels(f) fd <- fdroplevels(f) levd <- levels(fd) # Testing BY: expect_equal(attr(BY(d, f, sum), "row.names"), lev) expect_equal(dimnames(BY(m, f, sum))[[1L]], lev) expect_equal(names(BY(v, f, sum)), lev) # Fast Statistical Functions for(i in .FAST_STAT_FUN) { # print(i) FUN <- match.fun(i) expect_equal(attr(FUN(d, g = f), "row.names"), lev) expect_equal(dimnames(FUN(m, g = f))[[1L]], lev) expect_equal(names(FUN(v, g = f)), lev) expect_equal(attr(FUN(d, g = fd), "row.names"), levd) expect_equal(dimnames(FUN(m, g = fd))[[1L]], levd) expect_equal(names(FUN(v, g = fd)), levd) if(i != "fnobs") { expect_equal(attr(FUN(d, g = f, na.rm = FALSE), "row.names"), lev) expect_equal(dimnames(FUN(m, g = f, na.rm = FALSE))[[1L]], lev) expect_equal(names(FUN(v, g = f, na.rm = FALSE)), lev) } if(i %in% c("fsum", "fprod", "fmean", "fmedian", "fnth", "fmode", "fvar", "fsd")) { expect_equal(attr(FUN(d, g = f, w = w), "row.names"), lev) expect_equal(dimnames(FUN(m, g = f, w = w))[[1L]], lev) expect_equal(names(FUN(v, g = f, w = w)), lev) expect_equal(attr(FUN(d, g = f, w = w, na.rm = FALSE), "row.names"), lev) expect_equal(dimnames(FUN(m, g = f, w = w, na.rm = FALSE))[[1L]], lev) expect_equal(names(FUN(v, g = f, w = w, na.rm = FALSE)), lev) expect_equal(FUN(d, g = f, w = w), FUN(d, g = f)) expect_equal(FUN(m, g = f, w = w), FUN(m, g = f)) expect_equal(FUN(v, g = f, w = w), FUN(v, g = f)) } } # Other Statistical Functions for(i in setdiff(c(.FAST_FUN, .OPERATOR_FUN), .FAST_STAT_FUN)) { # print(i) FUN <- match.fun(i) if(grepl("hd", i, ignore.case = TRUE)) { expect_equal(FUN(d, fl = f), FUN(d, fl = fd)) expect_equal(FUN(m, fl = f), FUN(m, fl = fd)) expect_equal(FUN(v, fl = f), FUN(v, fl = fd)) expect_equal(FUN(d, fl = f, na.rm = FALSE), FUN(d, fl = fd, na.rm = FALSE)) expect_equal(FUN(m, fl = f, na.rm = FALSE), FUN(m, fl = fd, na.rm = FALSE)) expect_equal(FUN(v, fl = f, na.rm = FALSE), FUN(v, fl = fd, na.rm = FALSE)) expect_equal(FUN(d, fl = f, w = w), FUN(d, fl = fd)) expect_equal(FUN(m, fl = f, w = w), FUN(m, fl = fd)) expect_equal(FUN(v, fl = f, w = w), FUN(v, fl = fd)) expect_equal(FUN(d, fl = f, w = w, na.rm = FALSE), FUN(d, fl = fd, na.rm = FALSE)) expect_equal(FUN(m, fl = f, w = w, na.rm = FALSE), FUN(m, fl = fd, na.rm = FALSE)) expect_equal(FUN(v, fl = f, w = w, na.rm = FALSE), FUN(v, fl = fd, na.rm = FALSE)) } else { expect_equal(FUN(d, g = f), FUN(d, g = fd)) expect_equal(FUN(m, g = f), FUN(m, g = fd)) expect_equal(FUN(v, g = f), FUN(v, g = fd)) expect_equal(FUN(d, g = f, na.rm = FALSE), FUN(d, g = fd, na.rm = FALSE)) expect_equal(FUN(m, g = f, na.rm = FALSE), FUN(m, g = fd, na.rm = FALSE)) expect_equal(FUN(v, g = f, na.rm = FALSE), FUN(v, g = fd, na.rm = FALSE)) if(i %in% c("fscale", "STD", "fbetween", "B", "fwithin", "W")) { expect_equal(FUN(d, g = f, w = w), FUN(d, g = fd)) expect_equal(FUN(m, g = f, w = w), FUN(m, g = fd)) expect_equal(FUN(v, g = f, w = w), FUN(v, g = fd)) expect_equal(FUN(d, g = f, w = w, na.rm = FALSE), FUN(d, g = fd, na.rm = FALSE)) expect_equal(FUN(m, g = f, w = w, na.rm = FALSE), FUN(m, g = fd, na.rm = FALSE)) expect_equal(FUN(v, g = f, w = w, na.rm = FALSE), FUN(v, g = fd, na.rm = FALSE)) } } } }) test_that("Testing grouped_df methods", { skip_if_not_installed("magrittr") library(magrittr) for(sortg in c(TRUE, FALSE)) { for(retgrp in c(TRUE, FALSE)) { gdf <- wlddev %>% fsubset(year > 1990, region, income, PCGDP:ODA) %>% fgroup_by(region, income, return.groups = retgrp, sort = sortg) gdf[["wgt"]] <- round(abs(10*rnorm(fnrow(gdf))), 1) expect_visible(gdf %>% fmean) expect_visible(gdf %>% fmean(wgt)) expect_equal(gdf %>% fmean(wgt) %>% slt(-sum.wgt), gdf %>% fmean(wgt, keep.w = FALSE)) expect_visible(gdf %>% fmedian) expect_visible(gdf %>% fmedian(wgt)) expect_equal(gdf %>% fmedian(wgt) %>% slt(-sum.wgt), gdf %>% fmedian(wgt, keep.w = FALSE)) expect_visible(gdf %>% fnth) expect_visible(gdf %>% fnth(0.75)) expect_visible(gdf %>% fnth(0.75, wgt)) expect_equal(gdf %>% fnth(0.75, wgt) %>% slt(-sum.wgt), gdf %>% fnth(0.75, wgt, keep.w = FALSE)) expect_visible(gdf %>% fmode) expect_visible(gdf %>% fmode(wgt)) expect_equal(gdf %>% fmode(wgt) %>% slt(-sum.wgt), gdf %>% fmode(wgt, keep.w = FALSE)) expect_visible(gdf %>% fsum) expect_visible(gdf %>% fsum(wgt)) expect_equal(gdf %>% fsum(wgt) %>% slt(-sum.wgt), gdf %>% fsum(wgt, keep.w = FALSE)) expect_visible(gdf %>% fprod) expect_visible(gdf %>% fprod(wgt)) expect_equal(gdf %>% fprod(wgt) %>% slt(-prod.wgt), gdf %>% fprod(wgt, keep.w = FALSE)) expect_visible(gdf %>% fsd) expect_visible(gdf %>% fsd(wgt)) expect_equal(gdf %>% fsd(wgt) %>% slt(-sum.wgt), gdf %>% fsd(wgt, keep.w = FALSE)) expect_visible(gdf %>% fvar) expect_visible(gdf %>% fvar(wgt)) expect_equal(gdf %>% fvar(wgt) %>% slt(-sum.wgt), gdf %>% fvar(wgt, keep.w = FALSE)) expect_visible(gdf %>% fmin) expect_visible(gdf %>% fmax) expect_visible(gdf %>% ffirst) expect_visible(gdf %>% flast) expect_visible(gdf %>% fnobs) expect_visible(gdf %>% fndistinct) expect_visible(gdf %>% collapg) expect_visible(gdf %>% varying) expect_visible(gdf %>% varying(any_group = FALSE)) expect_visible(gdf %>% fmean(w = wgt)) # good? expect_equal(gdf %>% collapg(w = wgt) %>% slt(-wgt), gdf %>% collapg(w = wgt, keep.w = FALSE)) expect_visible(gdf %>% fscale) expect_visible(gdf %>% fscale(wgt)) expect_equal(gdf %>% fscale(wgt) %>% slt(-wgt), gdf %>% fscale(wgt, keep.w = FALSE)) expect_visible(gdf %>% STD) expect_visible(gdf %>% STD(wgt)) expect_equal(gdf %>% STD(wgt) %>% slt(-wgt), gdf %>% STD(wgt, keep.w = FALSE)) expect_equal(gdf %>% fscale, gdf %>% STD(stub = FALSE)) expect_visible(gdf %>% fbetween) expect_visible(gdf %>% fbetween(wgt)) expect_equal(gdf %>% fbetween(wgt) %>% slt(-wgt), gdf %>% fbetween(wgt, keep.w = FALSE)) expect_visible(gdf %>% B) expect_visible(gdf %>% B(wgt)) expect_equal(gdf %>% B(wgt) %>% slt(-wgt), gdf %>% B(wgt, keep.w = FALSE)) expect_equal(gdf %>% fbetween, gdf %>% B(stub = FALSE)) expect_visible(gdf %>% fwithin) expect_visible(gdf %>% fwithin(wgt)) expect_equal(gdf %>% fwithin(wgt) %>% slt(-wgt), gdf %>% fwithin(wgt, keep.w = FALSE)) expect_visible(gdf %>% W) expect_visible(gdf %>% W(wgt)) expect_equal(gdf %>% W(wgt) %>% slt(-wgt), gdf %>% W(wgt, keep.w = FALSE)) expect_equal(gdf %>% fwithin, gdf %>% W(stub = FALSE)) expect_visible(gdf %>% fcumsum) expect_visible(gdf %>% flag) expect_visible(gdf %>% L) expect_visible(gdf %>% F) expect_true(all_obj_equal(gdf %>% flag, gdf %>% L(stubs = FALSE), gdf %>% F(-1, stubs = FALSE))) expect_true(all_obj_equal(gdf %>% flag(-3:3), gdf %>% L(-3:3), gdf %>% F(3:-3))) expect_visible(gdf %>% fdiff) expect_visible(gdf %>% D) expect_true(all_obj_equal(gdf %>% fdiff, gdf %>% D(stubs = FALSE))) expect_equal(gdf %>% fdiff(-2:2, 1:2), gdf %>% D(-2:2, 1:2)) expect_visible(gdf %>% fdiff(rho = 0.95)) expect_visible(gdf %>% fdiff(-2:2, 1:2, rho = 0.95)) expect_visible(gdf %>% fdiff(log = TRUE)) expect_visible(gdf %>% fdiff(-2:2, 1:2, log = TRUE)) expect_visible(gdf %>% fdiff(log = TRUE, rho = 0.95)) expect_visible(gdf %>% fdiff(-2:2, 1:2, log = TRUE, rho = 0.95)) expect_visible(gdf %>% fgrowth) expect_visible(gdf %>% G) expect_true(all_obj_equal(gdf %>% fgrowth, gdf %>% G(stubs = FALSE))) expect_equal(gdf %>% fgrowth(-2:2, 1:2), gdf %>% G(-2:2, 1:2)) expect_visible(gdf %>% fgrowth(scale = 1)) expect_visible(gdf %>% fgrowth(-2:2, 1:2, scale = 1)) expect_visible(gdf %>% fgrowth(logdiff = TRUE)) expect_visible(gdf %>% fgrowth(-2:2, 1:2, logdiff = TRUE)) expect_visible(gdf %>% fgrowth(logdiff = TRUE, scale = 1)) expect_visible(gdf %>% fgrowth(-2:2, 1:2, logdiff = TRUE, scale = 1)) expect_equal(BY(gby(iris,Species), sum), BY(nv(gby(iris,Species)), sum)) } } }) # Also better not run on CRAN... test_that("0-length vectors give expected output", { funs <- .c(fsum, fprod, fmean, fmedian, fmin, fmax, fnth, fcumsum, fbetween, fwithin, fscale) for(i in funs) { FUN <- match.fun(i) if(i %!in% .c(fsum, fmin, fmax, fcumsum, fprod, fmean, fmedian, fnth)) { expect_true(all_identical(FUN(numeric(0)), FUN(integer(0)), numeric(0))) } else { expect_identical(FUN(numeric(0)), numeric(0)) if(i %in% .c(fmean, fprod, fnth, fmedian)) expect_identical(FUN(integer(0)), NA_real_) else expect_identical(FUN(integer(0)), integer(0)) } } funs <- .c(fmode, ffirst, flast) for(i in funs) { FUN <- match.fun(i) expect_identical(FUN(numeric(0)), numeric(0)) expect_identical(FUN(integer(0)), integer(0)) expect_identical(FUN(character(0)), character(0)) expect_identical(FUN(logical(0)), logical(0)) expect_identical(FUN(factor(0)), factor(0)) } funs <- .c(fvar, fsd) for(i in funs) { FUN <- match.fun(i) expect_identical(FUN(numeric(0)), NA_real_) expect_identical(FUN(integer(0)), NA_real_) } funs <- .c(fnobs, fndistinct) for(i in funs) { FUN <- match.fun(i) expect_identical(FUN(numeric(0)), 0L) expect_identical(FUN(integer(0)), 0L) } funs <- .c(flag, fdiff, fgrowth) for(i in funs) { FUN <- match.fun(i) expect_error(FUN(numeric(0))) expect_error(FUN(integer(0))) } funs <- .c(groupid, seqid) for(i in funs) { FUN <- match.fun(i) expect_identical(FUN(numeric(0)), integer(0)) expect_identical(FUN(integer(0)), integer(0)) } expect_identical(varying(numeric(0)), FALSE) expect_identical(TRA(numeric(0), 1), numeric(0)) }) } X <- matrix(rnorm(1000), ncol = 10) g <- qG(sample.int(10, 100, TRUE)) gf <- as_factor_qG(g) funs <- grep("hd|log", c(.FAST_FUN, .OPERATOR_FUN), ignore.case = TRUE, invert = TRUE, value = TRUE) test_that("functions work on plain matrices", { F <- getNamespace("collapse")$F for(i in funs) { expect_visible(match.fun(i)(X)) expect_visible(match.fun(i)(X, g = g)) expect_visible(match.fun(i)(X, g = gf)) expect_visible(match.fun(i)(X, g = g, use.g.names = FALSE)) expect_visible(match.fun(i)(X, g = gf, use.g.names = FALSE)) } }) Xl <- mctl(X) test_that("functions work on plain lists", { F <- getNamespace("collapse")$F for(i in funs) { expect_visible(match.fun(i)(Xl)) expect_visible(match.fun(i)(Xl, g = g, by = g)) expect_visible(match.fun(i)(Xl, g = gf, by = gf)) expect_visible(match.fun(i)(X, g = g, by = g, use.g.names = FALSE)) expect_visible(match.fun(i)(X, g = gf, by = gf, use.g.names = FALSE)) } }) test_that("time series functions work inside lm", { expect_equal(unname(coef(lm(mpg ~ L(cyl, 0:2), mtcars))), unname(coef(lm(mpg ~ cyl + L(cyl, 1) + L(cyl, 2), mtcars)))) expect_equal(unname(coef(lm(mpg ~ F(cyl, 0:2), mtcars))), unname(coef(lm(mpg ~ cyl + F(cyl, 1) + F(cyl, 2), mtcars)))) expect_equal(unname(coef(lm(mpg ~ D(cyl, 0:2), mtcars))), unname(coef(lm(mpg ~ cyl + D(cyl, 1) + D(cyl, 2), mtcars)))) expect_equal(unname(coef(lm(mpg ~ G(cyl, 0:2), mtcars))), unname(coef(lm(mpg ~ cyl + G(cyl, 1) + G(cyl, 2), mtcars)))) expect_equal(unname(coef(lm(mpg ~ L(L(cyl, 0:2)), mtcars))), unname(coef(lm(mpg ~ L(cyl) + L(cyl, 2) + L(cyl, 3), mtcars)))) expect_equal(unname(coef(lm(mpg ~ L(F(cyl, 0:2)), mtcars))), unname(coef(lm(mpg ~ L(cyl) + cyl + F(cyl, 1), mtcars)))) expect_equal(unname(coef(lm(mpg ~ L(D(cyl, 0:2)), mtcars))), unname(coef(lm(mpg ~ L(cyl) + L(D(cyl)) + L(D(cyl, 2)), mtcars)))) expect_equal(unname(coef(lm(mpg ~ L(G(cyl, 0:2)), mtcars))), unname(coef(lm(mpg ~ L(cyl) + L(G(cyl)) + L(G(cyl, 2)), mtcars)))) }) test_that("functions using welfords method properly deal with zero weights", { for(g in list(NULL, rep(1L, 3))) { expect_equal(unattrib(fvar(x = c(2, 1, 0), g = g, w = c(1, 1, 0), na.rm = TRUE)), 0.5) expect_equal(unattrib(fvar(x = c(2, 1, 3), g = g, w = c(0, 1, 1), na.rm = FALSE)), 2) expect_equal(unattrib(fsd(x = c(2, 1, 0), g = g, w = c(1, 1, 0), na.rm = TRUE)), sqrt(0.5)) expect_equal(unattrib(fsd(x = c(2, 1, 3), g = g, w = c(0, 1, 1), na.rm = FALSE)), sqrt(2)) expect_equal(unattrib(fscale(x = c(2, 1, 0), g = g, w = c(1, 1, 0), na.rm = TRUE)), (c(2, 1, 0)-1.5)/sqrt(0.5)) expect_equal(unattrib(fscale(x = c(2, 1, 3), g = g, w = c(0, 1, 1), na.rm = FALSE)), (c(2, 1, 3)-2)/sqrt(2)) expect_equal(unattrib(qsu(x = c(2, 1, 0), g = g, w = c(1, 1, 0))), c(2, 1.5, sqrt(0.5), 1, 2)) expect_equal(unattrib(qsu(x = c(2, 1, 3), g = g, w = c(0, 1, 1))), c(2, 2, sqrt(2), 1, 3)) expect_equal(unattrib(qsu(x = c(2, 1, 0), g = g, w = c(1, 1, 0), higher = TRUE))[1:5], c(2, 1.5, sqrt(0.5), 1, 2)) expect_equal(unattrib(qsu(x = c(2, 1, 3), g = g, w = c(0, 1, 1), higher = TRUE))[1:5], c(2, 2, sqrt(2), 1, 3)) } }) test_that("singleton groups are handled properly by all statistical functions", { w <- rep(1, fnrow(wlddev)) # Ordered g <- GRP(seq_row(wlddev), return.groups = FALSE) expect_equal(fmode(wlddev, g), wlddev) expect_equal(fmode(wlddev, g, w), wlddev) expect_equal(ffirst(wlddev, g), wlddev) expect_equal(flast(wlddev, g), wlddev) expect_equal(dapply(fndistinct(wlddev, g), unattrib), dapply(wlddev, function(x) as.integer(!is.na(x)))) expect_equal(fmode(wlddev, g, na.rm = FALSE), wlddev) expect_equal(fmode(wlddev, g, w, na.rm = FALSE), wlddev) expect_equal(ffirst(wlddev, g, na.rm = FALSE), wlddev) expect_equal(flast(wlddev, g, na.rm = FALSE), wlddev) expect_equal(dapply(fndistinct(wlddev, g, na.rm = FALSE), unattrib), dapply(wlddev, function(x) rep(1L, length(x)))) for(FUN in list(fmean, fmedian, fnth, fsum, fprod, fmin, fmax, fbetween, fcumsum)) { # print(FUN) expect_equal(FUN(nv(wlddev), g = g), nv(wlddev)) expect_equal(FUN(nv(wlddev), g = g, na.rm = FALSE), nv(wlddev)) expect_equal(FUN(nv(wlddev), g = g, w = w), nv(wlddev)) expect_equal(FUN(nv(wlddev), g = g, w = w, na.rm = FALSE), nv(wlddev)) } for(FUN in list(fvar, fsd, fscale, flag, fdiff, fgrowth)) { expect_true(all(dapply(FUN(nv(wlddev), g = g), allNA))) expect_true(all(dapply(FUN(nv(wlddev), g = g, na.rm = FALSE), allNA))) expect_true(all(dapply(FUN(nv(wlddev), g = g, w = w, n = -1), allNA))) expect_true(all(dapply(FUN(nv(wlddev), g = g, w = w, n = -1, na.rm = FALSE), allNA))) } # Unordered o <- radixorder(rnorm(fnrow(wlddev))) g <- GRP(o, return.groups = FALSE) wlduo <- setRownames(ss(wlddev, radixorder(o))) expect_equal(fmode(wlddev, g), wlduo) expect_equal(fmode(wlddev, g, w), wlduo) expect_equal(ffirst(wlddev, g), wlduo) expect_equal(flast(wlddev, g), wlduo) expect_equal(dapply(fndistinct(wlddev, g), unattrib), dapply(wlduo, function(x) as.integer(!is.na(x)))) expect_equal(fmode(wlddev, g, na.rm = FALSE), wlduo) expect_equal(fmode(wlddev, g, w, na.rm = FALSE), wlduo) expect_equal(ffirst(wlddev, g, na.rm = FALSE), wlduo) expect_equal(flast(wlddev, g, na.rm = FALSE), wlduo) expect_equal(dapply(fndistinct(wlddev, g, na.rm = FALSE), unattrib), dapply(wlduo, function(x) rep(1L, length(x)))) for(FUN in list(fmean, fmedian, fnth, fsum, fprod, fmin, fmax)) { # print(FUN) expect_equal(FUN(nv(wlddev), g = g), nv(wlduo)) expect_equal(FUN(nv(wlddev), g = g, na.rm = FALSE), nv(wlduo)) expect_equal(FUN(nv(wlddev), g = g, w = w), nv(wlduo)) expect_equal(FUN(nv(wlddev), g = g, w = w, na.rm = FALSE), nv(wlduo)) } for(FUN in list(fbetween, fcumsum)) { expect_equal(FUN(nv(wlddev), g), nv(wlddev)) expect_equal(FUN(nv(wlddev), g, na.rm = FALSE), nv(wlddev)) expect_equal(FUN(nv(wlddev), g, w), nv(wlddev)) expect_equal(FUN(nv(wlddev), g, w, na.rm = FALSE), nv(wlddev)) } for(FUN in list(fvar, fsd, fscale, flag, fdiff, fgrowth)) { expect_true(all(dapply(FUN(nv(wlddev), g = g), allNA))) expect_true(all(dapply(FUN(nv(wlddev), g = g, na.rm = FALSE), allNA))) expect_true(all(dapply(FUN(nv(wlddev), g = g, w = w, n = -1), allNA))) expect_true(all(dapply(FUN(nv(wlddev), g = g, w = w, n = -1, na.rm = FALSE), allNA))) } }) test_that("functions work for data frames with zero rows", { mtc0 <- qDF(mtcars)[NULL, ] expect_equal(mtc0, funique(mtc0)) expect_equal(mtc0, funique(mtc0, sort = TRUE)) expect_equal(mtc0, roworderv(mtc0)) expect_visible(colorder(mtc0, mpg, hp)) expect_visible(GRP(mtc0)) expect_visible(fgroup_by(mtc0, cyl, vs, am)) expect_visible(GRP(mtc0, sort = FALSE)) expect_visible(fgroup_by(mtc0, cyl, vs, am, sort = FALSE)) expect_visible(fduplicated(mtc0)) expect_false(any_duplicated(mtc0)) expect_visible(fselect(mtc0, hp, carb)) expect_visible(get_vars(mtc0, 9:8)) }) test_that("issue with integer followed by NA #432", { for (f in setdiff(.FAST_STAT_FUN, c("fvar", "fsd", "fnobs", "fndistinct"))) { # if(!isTRUE(all.equal(match.fun(f)(c(10L, NA)), 10L))) print(f) expect_equal(match.fun(f)(c(10L, NA)), 10L) expect_equal(match.fun(f)(c(NA, 10L)), 10L) expect_equal(match.fun(f)(c(10, NA)), 10) expect_equal(match.fun(f)(c(NA, 10)), 10) expect_equal(match.fun(f)(c(10L, NA), g = rep(1L, 2), use.g.names = FALSE), 10L) expect_equal(match.fun(f)(c(NA, 10L), g = rep(1L, 2), use.g.names = FALSE), 10L) expect_equal(match.fun(f)(c(10, NA), g = rep(1L, 2), use.g.names = FALSE), 10) expect_equal(match.fun(f)(c(NA, 10), g = rep(1L, 2), use.g.names = FALSE), 10) # na.rm = FALSE if(f %!in% c("fmode", "ffirst")) expect_equal(match.fun(f)(c(10L, NA), na.rm = FALSE), NA_integer_) if(f != "flast") expect_equal(match.fun(f)(c(NA, 10L), na.rm = FALSE), NA_integer_) if(f %!in% c("fmode", "ffirst")) expect_equal(match.fun(f)(c(10, NA), na.rm = FALSE), NA_real_) if(f != "flast") expect_equal(match.fun(f)(c(NA, 10), na.rm = FALSE), NA_real_) # Some functions are optimized and don't check here # expect_equal(match.fun(f)(c(10L, NA), g = rep(1L, 2), na.rm = FALSE, use.g.names = FALSE), NA_integer_) # expect_equal(match.fun(f)(c(NA, 10L), g = rep(1L, 2), na.rm = FALSE, use.g.names = FALSE), NA_integer_) if(f %!in% c("fmode", "ffirst")) expect_equal(match.fun(f)(c(10, NA), g = rep(1L, 2), na.rm = FALSE, use.g.names = FALSE), NA_real_) if(f != "flast") expect_equal(match.fun(f)(c(NA, 10), g = rep(1L, 2), na.rm = FALSE, use.g.names = FALSE), NA_real_) } skip_if_not(Sys.getenv("OMP") == "TRUE") for (f in c("fsum", "fmean", "fmode", "fnth", "fmedian")) { expect_equal(match.fun(f)(c(10L, rep(NA_integer_, 1e5)), nthreads = 2L), 10L) expect_equal(match.fun(f)(c(rep(NA_integer_, 1e5), 10L), nthreads = 2L), 10L) expect_equal(match.fun(f)(c(10, rep(NA_real_, 1e5)), nthreads = 2L), 10) expect_equal(match.fun(f)(c(rep(NA_real_, 1e5), 10), nthreads = 2L), 10) expect_equal(match.fun(f)(c(10L, rep(NA_integer_, 1e5)), g = rep(1L, 1e5+1), nthreads = 2L, use.g.names = FALSE), 10L) expect_equal(match.fun(f)(c(rep(NA_integer_, 1e5), 10L), g = rep(1L, 1e5+1), nthreads = 2L, use.g.names = FALSE), 10L) expect_equal(match.fun(f)(c(10, rep(NA_real_, 1e5)), g = rep(1L, 1e5+1), nthreads = 2L, use.g.names = FALSE), 10) expect_equal(match.fun(f)(c(rep(NA_real_, 1e5), 10), g = rep(1L, 1e5+1), nthreads = 2L, use.g.names = FALSE), 10) # na.rm = FALSE expect_equal(match.fun(f)(c(10L, rep(NA_integer_, 1e5)), na.rm = FALSE, nthreads = 2L), NA_integer_) expect_equal(match.fun(f)(c(rep(NA_integer_, 1e5), 10L), na.rm = FALSE, nthreads = 2L), NA_integer_) expect_equal(match.fun(f)(c(10, rep(NA_real_, 1e5)), na.rm = FALSE, nthreads = 2L), NA_real_) expect_equal(match.fun(f)(c(rep(NA_real_, 1e5), 10), na.rm = FALSE, nthreads = 2L), NA_real_) # Some functions are optimized and don't check here # expect_equal(match.fun(f)(c(10L, rep(NA_integer_, 1e5)), g = rep(1L, 1e5+1), na.rm = FALSE, nthreads = 2L, use.g.names = FALSE), NA_integer_) # expect_equal(match.fun(f)(c(rep(NA_integer_, 1e5), 10L), g = rep(1L, 1e5+1), na.rm = FALSE, nthreads = 2L, use.g.names = FALSE), NA_integer_) expect_equal(match.fun(f)(c(10, rep(NA_real_, 1e5)), g = rep(1L, 1e5+1), na.rm = FALSE, nthreads = 2L, use.g.names = FALSE), NA_real_) expect_equal(match.fun(f)(c(rep(NA_real_, 1e5), 10), g = rep(1L, 1e5+1), na.rm = FALSE, nthreads = 2L, use.g.names = FALSE), NA_real_) } }) test_that("fmedian ties handled properly with weights", { x <- c(1, 2, 3, 4) w <- c(2.5, 2.4, 3.8, 1.1) expect_equal(c(fmedian(x, w = w, ties = "mean"), fmedian(x, w = w, ties = "min"), fmedian(x, w = w, ties = "max")), c(2.5, 2, 3)) w <- c(2.5, 2.4, 3.7, 1.2) expect_equal(c(fmedian(x, w = w, ties = "mean"), fmedian(x, w = w, ties = "min"), fmedian(x, w = w, ties = "max")), c(2.5, 2, 3)) }) options(warn = 1)