tests/testthat/test_any_firstafter.R

In datacamp/funneljoin: Time-Based Joins to Analyze Sequences of Events

context("any-firstafter joining")
library(dplyr)

test_that("after_join works with mode = inner and type = any-firstafter", {

  res <- after_join(landed, registered, by_user = "user_id", by_time = c("timestamp" = "timestamp"), mode = "inner", type = "any-firstafter")

  expect_is(res, "tbl_df")
  expect_equal(names(res), c("user_id", "timestamp.x", "timestamp.y"))
  expect_gte(nrow(landed), nrow(res))
  expect_true(all(res$timestamp.y >= res$timestamp.x))
  expect_true(4 %in% res$user_id)
  expect_true(1 %in% res$user_id)
  expect_true(!7 %in% res$user_id)
  expect_true(!2 %in% res$user_id)
  expect_true(all(!is.na(res$timestamp.x)))
  expect_true(all(!is.na(res$timestamp.y)))
  expect_true(all(!is.na(res$user_id)))
})

test_that("after_join works with mode = left and type = any-firstafter", {

  res <- after_join(landed, registered, by_user = "user_id", by_time = c("timestamp" = "timestamp"), mode = "left", type = "any-firstafter")

  expect_is(res, "tbl_df")
  expect_equal(names(res), c("user_id", "timestamp.x", "timestamp.y"))
  expect_true(all(res$timestamp.y >= res$timestamp.x |
                    is.na(res$timestamp.y)))
  expect_gte(nrow(landed), nrow(res))
  expect_true(4 %in% res$user_id)
  expect_true(1 %in% res$user_id)
  expect_true(all(!is.na(res$timestamp.x)))
  expect_true(any(is.na(res$timestamp.y)))
  expect_true(all(!is.na(res$user_id)))
})

test_that("after_join works with mode = right and type = any-firstafter", {

  res <- after_join(landed, registered, by_user = "user_id", by_time = c("timestamp" = "timestamp"), mode = "right", type = "any-firstafter")

  expect_is(res, "tbl_df")
  expect_equal(names(res), c("user_id", "timestamp.x", "timestamp.y"))
  expect_true(all(res$timestamp.y >= res$timestamp.x |
                    is.na(res$timestamp.x)))
  expect_true(4 %in% res$user_id)
  expect_true(1 %in% res$user_id)
  expect_true(!2 %in% res$user_id)
  expect_true(all(!is.na(res$timestamp.y)))
  expect_true(any(is.na(res$timestamp.x)))
  expect_true(all(!is.na(res$user_id)))
})

test_that("after_join works with mode = anti and type = any-firstafter", {

  res <- after_join(landed, registered, by_user = "user_id", by_time = c("timestamp" = "timestamp"), mode = "anti", type = "any-firstafter")

  expect_is(res, "tbl_df")
  expect_equal(names(res), c("user_id", "timestamp"))
  expect_true(2 %in% res$user_id)
  expect_true(!3 %in% res$user_id)
  expect_true(all(!is.na(res$timestamp)))
  expect_true(all(!is.na(res$user_id)))
  expect_true(as.Date("2018-07-01") %in% res$timestamp)
})

test_that("after_join works with mode = semi and type = any-firstafter", {

  res <- after_join(landed, registered, by_user = "user_id", by_time = c("timestamp" = "timestamp"), mode = "semi", type = "any-firstafter")

  expect_is(res, "tbl_df")
  expect_equal(names(res), c("user_id", "timestamp"))
  expect_true(!2 %in% res$user_id)
  expect_true(3 %in% res$user_id)
  expect_true(all(!is.na(res$timestamp)))
  expect_true(all(!is.na(res$user_id)))
  expect_true(as.Date("2018-07-10") %in% res$timestamp)
  expect_equal(nrow(filter(res, user_id == 6)), 2)
})

test_that("after_join works with mode = full and type = any-firstafter", {

  res <- after_join(landed, registered, by_user = "user_id", by_time = c("timestamp" = "timestamp"), mode = "full", type = "any-firstafter")

  expect_is(res, "tbl_df")
  expect_equal(names(res), c("user_id", "timestamp.x", "timestamp.y"))
  expect_true(nrow(res) >= 8)
  expect_true(all(res$timestamp.y >= res$timestamp.x |
                    is.na(res$timestamp.x) |
                    is.na(res$timestamp.y)))
  expect_gt(nrow(res), dplyr::n_distinct(registered$user_id))
  expect_gt(nrow(res), dplyr::n_distinct(landed$user_id))
  expect_true(any(is.na(res$timestamp.x)))
  expect_true(any(is.na(res$timestamp.y)))
  expect_true(all(!is.na(res$user_id)))
  expect_equal(nrow(filter(res, user_id == 6)), 3)
})