inst/tinytest/test_mcf.formula.R

In reda: Recurrent Event Data Analysis

library(reda)

data(valveSeats)
valveSeats$group_ <- cut(valveSeats$ID, c(250, 400, 450))

## test compatibility of 'Survr'
expect_warning(mcf(Survr(ID, Days, No.) ~ 1, valveSeats),
               "deprecated")

## test na.action
testDat <- valveSeats
testDat[1, "group_"] <- NA
test_mcf <- mcf(Recur(Days, ID, No.) ~ group_,
                testDat, na.action = "na.exclude",
                variance = "bootstrap",
                control = list(B = 30))
expect_equal(nrow(valveSeats) - nrow(test_mcf@data), 1L)

## point estimates only without se and ci estimates
mcf_res1 <- mcf(Recur(time, ID, event) ~ 1, data = simuDat, variance = "none")
expect_equivalent(rep(NA_real_, nrow(mcf_res1@MCF)), mcf_res1@MCF$se)

## test bootstrap variance
expect_equivalent(class(
    mcf(Recur(Days, ID, No.) ~ group_, valveSeats,
        variance = "bootstrap", logConfInt = TRUE,
        control = list(B = 30, se.method = "normality"))
), "mcf.formula")
expect_equivalent(class(
    mcf(Recur(Days, ID, No.) ~ group_, valveSeats,
        variance = "bootstrap",
        control = list(B = 30, se.method = "normality",
                       ci.method = "percentile"))
), "mcf.formula")
expect_equivalent(class(
    mcf(Recur(Days, ID, No.) ~ group_, valveSeats,
        variance = "bootstrap", logConfInt = TRUE,
        control = list(B = 30, se.method = "normality",
                       ci.method = "percentile"))
), "mcf.formula")

## test its plot-method
mcf0 <- mcf(Recur(Days, ID, No.) ~ 1, valveSeats)
expect_equivalent(class(
    plot(mcf0, conf.int = TRUE, addOrigin = TRUE, mark.time = TRUE)
), c("gg", "ggplot"))
expect_equivalent(class(
    plot(test_mcf, conf.int = TRUE, addOrigin = TRUE, mark.time = TRUE,
         legendLevels = c("G1", "G2"), legendName = "Group")
), c("gg", "ggplot"))
expect_equivalent(class(
    plot(test_mcf, lty = 1:2, col = 1:2)
), c("gg", "ggplot"))

## show method
expect_equal(show(mcf0), mcf0)