diff --git a/R/challengeR.R b/R/challengeR.R
index 6d9e9c2..45d53d0 100644
--- a/R/challengeR.R
+++ b/R/challengeR.R
@@ -1,207 +1,207 @@
# Copyright (c) German Cancer Research Center (DKFZ)
# All rights reserved.
#
# This file is part of challengeR.
#
# challengeR is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 2 of the License, or
# (at your option) any later version.
#
# challengeR is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with challengeR. If not, see <https://www.gnu.org/licenses/>.
#' Constructs a challenge object
#'
#' Constructs an S3 object to represent the configuration of an assessment data set originating from a benchmarking competition (so-called "challenge").
#'
#' @section Assessment data set:
#' The toolkit provides visualization approaches for both challenges designed around a single task (single-task challenges) and for challenges comprising multiple tasks (multi-task challenges).
#' For a single-task challenge, the assessment data set (argument \code{object}) requires the following columns:
#' \itemize{
#' \item test case identifier (string or numeric)
#' \item algorithm identifier (string or numeric)
#' \item performance value (numeric)
#' }
#'
#' For a multi-task challenge, the assessment data set (argument \code{object}) requires the following columns:
#' \itemize{
#' \item task identifier (string or numeric)
#' \item test case identifier (string or numeric)
#' \item algorithm identifier (string or numeric)
#' \item performance value (numeric)
#' }
#'
#' @section Sanity check:
#' It is highly recommended that the sanity check is not disabled when the data set is provided initially.
#' It checks that:
#' \itemize{
#' \item performance values are numeric (if not, raises error)
#' \item algorithm performances are observed for all cases (if not, adds them as NA and emits a message)
#' \item cases appear only once for the same algorithm (if not, raises error)
#' }
#' If the argument \code{na.treat} for treatment of NA is specified, NAs will be handled respectively.
#'
#' It might be reasonable to disable the sanity check for further computations (e.g., for performance reasons
#' during bootstrapping (\code{\link{bootstrap.ranked.list}}) where cases are actually allowed to appear more than once for the same algorithm).
#'
#' @param object A data frame containing the assessment data.
#' @param case A string specifying the name of the column that contains the case identifiers.
#' @param algorithm A string specifying the name of the column that contains the algorithm identifiers.
#' @param value A string specifying the name of the column that contains the performance values.
#' @param by A string specifying the name of the column that contains the task identifiers. Required for multi-task data set.
#' @param taskName A string specifying the task name for single-task data set that does not contain a task column.
#' This argument is optional for a single-task data set and is ignored for a multi-task data set.
#' @param annotator If multiple annotators annotated the test cases, a string specifying the name of the column that contains the annotator identifiers. Only applies to rang-then-aggregate. Use with caution: Currently not tested.
#' @param smallBetter A boolean specifying whether small performance values indicate better algorithm performance.
#' @param na.treat Indicates how missing perfomance values are treated if sanity check is enabled. It can be 'na.rm', numeric value or function.
#' For a numeric value or function, NAs will be replaced by the specified values. For 'na.rm', rows that contain missing values will be removed.
#' @param check A boolean to indicate to perform a sanity check of the specified data set and arguments if set to \code{TRUE}.
#'
#' @return An S3 object to represent the configuration of an assessment data set.
#'
#' @examples
#' # single-task data set
#'
#' # multi-task data set
#'
#' @export
as.challenge=function(object,
case,
algorithm,
value,
by=NULL,
taskName=NULL,
annotator=NULL,
smallBetter=FALSE,
na.treat=NULL, # optional
check=TRUE) {
object=as.data.frame(object[,c(value, algorithm, case, by, annotator)])
-
+ object[[algorithm]] <- as.factor(object[[algorithm]])
# sanity checks
if (check) {
if (!is.null(by) && !is.null(taskName)) {
warning("Argument 'taskName' is ignored for multi-task data set.")
}
# Add task column for data set without task column by using the specified task name.
if (is.null(by) && !is.null(taskName)) {
taskName <- trimws(taskName)
if (taskName == "") {
stop("Argument 'taskName' is empty.")
}
object <- cbind(task=taskName, object)
by = "task"
}
# Add task column for data set without task column by using a dummy task name.
if (is.null(by) && is.null(taskName)) {
object <- cbind(task="dummyTask", object)
by = "task"
}
object=splitby(object,by=by)
object=lapply(object,droplevels)
missingData = n.missing = list()
for (task in names(object)) {
if (!all(is.numeric(object[[task]][[value]]))) stop("Performance values must be numeric.")
n.missing[[task]] <- sum(is.na(object[[task]][[value]])) # already missing before na.treat; for report
if (n.missing[[task]]>0) message("Note: ", n.missing, " missing cases have been found in the data set.")
# check for missing cases
missingData[[task]]=object[[task]] %>%
expand(!!as.symbol(algorithm),
!!as.symbol(case))%>%
anti_join(object[[task]],
by=c( algorithm,case))
if (nrow(missingData[[task]])>0) {
if (length(object) == 1 ) { # single task
message("Performance of not all algorithms has been observed for all cases. Therefore, missings have been inserted in the following cases:")
} else { # multi task
message("Performance of not all algorithms has been observed for all cases in task '",
task,
"'. Therefore, missings have been inserted in the following cases:")
}
print(as.data.frame(missingData[[task]]))
object[[task]]=as.data.frame(object[[task]] %>%
complete(!!as.symbol(by),
!!as.symbol(algorithm),
!!as.symbol(case)))
}
# check duplicate cases
all1=apply(table(object[[task]][[algorithm]],
object[[task]][[case]]),
2,
function(x) all(x==1))
if (!all(all1)) {
n.duplicated <- sum(all1!=1)
if (length(object) == 1 ) { # single task
if (n.duplicated/length(all1) >= 1/5) { # at least a quarter of the cases is duplicated
stop ("The following case(s) appear(s) more than once for the same algorithm. Please revise. ",
"Or are you considering a multi-task challenge and forgot to specify argument 'by'?\n",
"Case(s): ",
paste(names(which(all1!=1)), collapse=", ")
)
} else {
stop ("The following case(s) appear(s) more than once for the same algorithm. Please revise.\n",
"Case(s): ",
paste(names(which(all1!=1)), collapse=", ")
)
}
} else { # multi task
stop ("The following case(s) appear(s) more than once for the same algorithm in task '",
task, "'. Please revise.\n",
"Case(s): ",
paste(names(which(all1!=1)), collapse=", ")
)
}
}
if (!is.null(na.treat)) {
if (is.numeric(na.treat)) object[[task]][,value][is.na(object[[task]][,value])]=na.treat
else if (is.function(na.treat)) object[[task]][,value][is.na(object[[task]][,value])]=na.treat(object[[task]][,value][is.na(object[[task]][,value])])
else if (is.character(na.treat) && na.treat=="na.rm") object[[task]]=object[[task]][!is.na(object[[task]][,value]),]
}
}
}
if (check==TRUE && (any(sapply(missingData, function(x) nrow(x))>0) |any(n.missing>0))) {
if (is.null(na.treat)) message("For aggregate-then-rank, na.treat will have to be specified. ",
"For rank-then-aggregate, missings will implicitly lead to the algorithm ranked last for the missing test case.",
"na.treat obligatory if report is intended to be compiled."
)
else if (is.numeric(na.treat)) message("All missings have been replaced by the value ", na.treat,".\n")
else if (is.character(na.treat) && na.treat=="na.rm") message("All missings have been removed.")
else if (is.function(na.treat)) {
message("Missings have been replaced using function ")
print(na.treat)
}
}
if (check==TRUE){
attr(object,"n.missing")=n.missing
attr(object,"missingData")=missingData
}
attr(object,"na.treat")=na.treat
attr(object,"algorithm")=algorithm
attr(object,"value")=value
attr(object,"case")=case
attr(object,"annotator")=annotator
attr(object,"by")=by
attr(object,"smallBetter")=smallBetter
attr(object,"check")=check
class(object)=c("challenge", class(object))
object
}
diff --git a/tests/testthat/test-challenge.R b/tests/testthat/test-challenge.R
index b3c7ffe..819d1c5 100644
--- a/tests/testthat/test-challenge.R
+++ b/tests/testthat/test-challenge.R
@@ -1,596 +1,633 @@
# Copyright (c) German Cancer Research Center (DKFZ)
# All rights reserved.
#
# This file is part of challengeR.
#
# challengeR is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 2 of the License, or
# (at your option) any later version.
#
# challengeR is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with challengeR. If not, see <https://www.gnu.org/licenses/>.
test_that("empty attribute 'taskName' raises error for single-task challenge", {
data <- rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C1"))
expect_error(as.challenge(data, taskName="", algorithm="algo", case="case", value="value", smallBetter=FALSE),
"Argument 'taskName' is empty.", fixed=TRUE)
})
test_that("only whitespaces in attribute 'taskName' raises error for single-task challenge", {
data <- rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C1"))
expect_error(as.challenge(data, taskName=" ", algorithm="algo", case="case", value="value", smallBetter=FALSE),
"Argument 'taskName' is empty.", fixed=TRUE)
})
test_that("attributes are set for single-task challenge with specified task name", {
data <- rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C1"))
actualChallenge <- as.challenge(data, taskName="T1", algorithm="algo", case="case", value="value", smallBetter=FALSE)
expect_equal(attr(actualChallenge, "annotator"), NULL)
expect_equal(attr(actualChallenge, "by"), "task")
expect_equal(attr(actualChallenge, "smallBetter"), FALSE)
expect_equal(attr(actualChallenge, "check"), TRUE)
expect_equal(as.vector(actualChallenge$T1$algo), c("A1", "A2"))
expect_equal(as.vector(actualChallenge$T1$value), c(0.8, 0.6))
expect_equal(as.vector(actualChallenge$T1$case), c("C1", "C1"))
expect_equal(as.vector(actualChallenge$T1$task), c("T1", "T1"))
# expect that there's no attribute "task"
expect_equal(attr(actualChallenge, "task"), NULL)
expect_equal(attr(actualChallenge$T1, "task"), NULL)
expect_equal(attr(actualChallenge$T2, "task"), NULL)
})
test_that("attributes are set for single-task challenge with dummy task name", {
data <- rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C1"))
actualChallenge <- as.challenge(data, algorithm="algo", case="case", value="value", smallBetter=FALSE)
expect_equal(attr(actualChallenge, "annotator"), NULL)
expect_equal(attr(actualChallenge, "by"), "task")
expect_equal(attr(actualChallenge, "smallBetter"), FALSE)
expect_equal(attr(actualChallenge, "check"), TRUE)
expect_equal(as.vector(actualChallenge$dummyTask$algo), c("A1", "A2"))
expect_equal(as.vector(actualChallenge$dummyTask$value), c(0.8, 0.6))
expect_equal(as.vector(actualChallenge$dummyTask$case), c("C1", "C1"))
expect_equal(as.vector(actualChallenge$dummyTask$task), c("dummyTask", "dummyTask"))
# expect that there's no attribute "task"
expect_equal(attr(actualChallenge, "task"), NULL)
expect_equal(attr(actualChallenge$dummyTask, "task"), NULL)
expect_equal(attr(actualChallenge$dummyTask, "task"), NULL)
})
test_that("leading and trailing whitespaces are trimmed for attribute 'taskName'", {
data <- rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C1"))
actualChallenge <- as.challenge(data, taskName=" T1 ", algorithm="algo", case="case", value="value", smallBetter=FALSE)
expect_equal(as.vector(actualChallenge$T1$algo), c("A1", "A2"))
})
test_that("attributes are set for multi-task challenge", {
dataTask1 <- cbind(task="T1",
rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C1")
))
dataTask2 <- cbind(task="T2",
rbind(
data.frame(algo="A1", value=0.2, case="C1"),
data.frame(algo="A2", value=0.3, case="C1")
))
data <- rbind(dataTask1, dataTask2)
actualChallenge <- as.challenge(data, by="task", algorithm="algo", case="case", value="value", smallBetter=TRUE)
expect_equal(attr(actualChallenge, "annotator"), NULL)
expect_equal(attr(actualChallenge, "by"), "task")
expect_equal(attr(actualChallenge, "smallBetter"), TRUE)
expect_equal(attr(actualChallenge, "check"), TRUE)
expect_equal(as.vector(actualChallenge$T1$algo), c("A1", "A2"))
expect_equal(as.vector(actualChallenge$T1$value), c(0.8, 0.6))
expect_equal(as.vector(actualChallenge$T1$case), c("C1", "C1"))
expect_equal(as.vector(actualChallenge$T1$task), c("T1", "T1"))
expect_equal(as.vector(actualChallenge$T2$algo), c("A1", "A2"))
expect_equal(as.vector(actualChallenge$T2$value), c(0.2, 0.3))
expect_equal(as.vector(actualChallenge$T2$case), c("C1", "C1"))
expect_equal(as.vector(actualChallenge$T2$task), c("T2", "T2"))
# expect that there's no attribute "task"
expect_equal(attr(actualChallenge, "task"), NULL)
expect_equal(attr(actualChallenge$T1, "task"), NULL)
expect_equal(attr(actualChallenge$T2, "task"), NULL)
})
test_that("attributes are set for multi-task challenge with sanity check disabled", {
dataTask1 <- cbind(task="T1",
rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C1")
))
dataTask2 <- cbind(task="T2",
rbind(
data.frame(algo="A1", value=0.2, case="C1"),
data.frame(algo="A2", value=0.3, case="C1")
))
data <- rbind(dataTask1, dataTask2)
actualChallenge <- as.challenge(data, by="task", algorithm="algo", case="case", value="value", smallBetter=TRUE, check=FALSE)
expect_equal(attr(actualChallenge, "annotator"), NULL)
expect_equal(attr(actualChallenge, "by"), "task")
expect_equal(attr(actualChallenge, "smallBetter"), TRUE)
expect_equal(attr(actualChallenge, "check"), FALSE)
expect_equal(as.vector(actualChallenge$algo), c("A1", "A2", "A1", "A2"))
expect_equal(as.vector(actualChallenge$value), c(0.8, 0.6, 0.2, 0.3))
expect_equal(as.vector(actualChallenge$case), c("C1", "C1", "C1", "C1"))
expect_equal(as.vector(actualChallenge$task), c("T1", "T1", "T2", "T2"))
})
test_that("attribute 'taskName' is ignored for multi-task challenge", {
dataTask1 <- cbind(task="T1",
rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C1")
))
dataTask2 <- cbind(task="T2",
rbind(
data.frame(algo="A1", value=0.2, case="C1"),
data.frame(algo="A2", value=0.3, case="C1")
))
data <- rbind(dataTask1, dataTask2)
expect_warning(as.challenge(data, taskName="T1", by="task", algorithm="algo", case="case", value="value", smallBetter=TRUE),
"Argument 'taskName' is ignored for multi-task data set.", fixed=TRUE)
})
test_that("missing algorithm performances are added as NAs for single-task challenge", {
data <- rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C2"))
expect_message(actualChallenge <- as.challenge(data, taskName="T1", algorithm="algo", case="case", value="value", smallBetter=FALSE),
"Performance of not all algorithms has been observed for all cases. Therefore, missings have been inserted in the following cases:", fixed=TRUE)
expect_equal(as.vector(actualChallenge$T1$algo), c("A1", "A1", "A2", "A2"))
expect_equal(as.vector(actualChallenge$T1$value), c(0.8, NA, NA, 0.6))
expect_equal(as.vector(actualChallenge$T1$case), c("C1", "C2", "C1", "C2"))
expect_equal(as.vector(actualChallenge$T1$task), c("T1", "T1", "T1", "T1"))
})
test_that("multi-task data set containing one task is interpreted as single-task data set, missing algorithm performances are added", {
data <- cbind(task="T1",
rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C2")
))
# do not specify parameter "by" to interpret multi-task data set as single-task data set
expect_message(actualChallenge <- as.challenge(data, taskName="T1", algorithm="algo", case="case", value="value", smallBetter=FALSE),
"Performance of not all algorithms has been observed for all cases. Therefore, missings have been inserted in the following cases:", fixed=TRUE)
expect_equal(as.vector(actualChallenge$T1$algo), c("A1", "A1", "A2", "A2"))
expect_equal(as.vector(actualChallenge$T1$value), c(0.8, NA, NA, 0.6))
expect_equal(as.vector(actualChallenge$T1$case), c("C1", "C2", "C1", "C2"))
})
test_that("missing algorithm performances are added as NAs for multi-task challenge (2 tasks in data set)", {
dataTask1 <- cbind(task="T1",
rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C2")
))
dataTask2 <- cbind(task="T2",
rbind(
data.frame(algo="A1", value=0.2, case="C1"),
data.frame(algo="A1", value=0.3, case="C2"),
data.frame(algo="A2", value=0.4, case="C1")
))
data <- rbind(dataTask1, dataTask2)
expect_message(actualChallenge <- as.challenge(data, by="task", algorithm="algo", case="case", value="value", smallBetter=FALSE),
"Performance of not all algorithms has been observed for all cases in task 'T1'. Therefore, missings have been inserted in the following cases:", fixed=TRUE)
expect_equal(as.vector(actualChallenge$T1$algo), c("A1", "A1", "A2", "A2"))
expect_equal(as.vector(actualChallenge$T1$value), c(0.8, NA, NA, 0.6))
expect_equal(as.vector(actualChallenge$T1$case), c("C1", "C2", "C1", "C2"))
expect_equal(as.vector(actualChallenge$T1$task), c("T1", "T1", "T1", "T1"))
expect_equal(as.vector(actualChallenge$T2$algo), c("A1", "A1", "A2", "A2"))
expect_equal(as.vector(actualChallenge$T2$value), c(0.2, 0.3, 0.4, NA))
expect_equal(as.vector(actualChallenge$T2$case), c("C1", "C2", "C1", "C2"))
expect_equal(as.vector(actualChallenge$T2$task), c("T2", "T2", "T2", "T2"))
})
test_that("missing algorithm performances are not added as NA with sanity check disabled for single-task challenge", {
data <- rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C2"))
actualChallenge <- as.challenge(data, taskName="T1", algorithm="algo", case="case", value="value", smallBetter=FALSE, check=FALSE)
expect_equal(as.vector(actualChallenge$algo), c("A1", "A2"))
expect_equal(as.vector(actualChallenge$value), c(0.8, 0.6))
expect_equal(as.vector(actualChallenge$case), c("C1", "C2"))
})
test_that("missing algorithm performances are not added as NA with sanity check disabled for multi-task challenge (2 tasks in data set)", {
dataTask1 <- cbind(task="T1",
rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C2")
))
dataTask2 <- cbind(task="T2",
rbind(
data.frame(algo="A1", value=0.2, case="C1"),
data.frame(algo="A1", value=0.3, case="C2"),
data.frame(algo="A2", value=0.4, case="C1")
))
data <- rbind(dataTask1, dataTask2)
actualChallenge <- as.challenge(data, by="task", algorithm="algo", case="case", value="value", smallBetter=FALSE, check=FALSE)
expect_equal(as.vector(actualChallenge$algo), c("A1", "A2", "A1", "A1", "A2"))
expect_equal(as.vector(actualChallenge$value), c(0.8, 0.6, 0.2, 0.3, 0.4))
expect_equal(as.vector(actualChallenge$case), c("C1", "C2", "C1", "C2", "C1"))
})
test_that("case cannot appear more than once per algorithm for single-task challenge", {
data <- rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A1", value=0.8, case="C1"))
expect_error(as.challenge(data, taskName="T1", algorithm="algo", case="case", value="value", smallBetter=FALSE),
"The following case(s) appear(s) more than once for the same algorithm. Please revise. Or are you considering a multi-task challenge and forgot to specify argument 'by'?\nCase(s): C1", fixed=TRUE)
})
test_that("multi-task data set containing one task is interpreted as single-task data set, case cannot appear more than once per algorithm", {
data <- cbind(task="T1",
rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A1", value=0.8, case="C1")
))
# do not specify parameter "by" to interpret multi-task data set as single-task data set
expect_error(as.challenge(data, taskName="T1", algorithm="algo", case="case", value="value", smallBetter=FALSE),
"The following case(s) appear(s) more than once for the same algorithm. Please revise. Or are you considering a multi-task challenge and forgot to specify argument 'by'?\nCase(s): C1", fixed=TRUE)
})
test_that("case cannot appear more than once per algorithm for multi-task challenge (1 task in data set)", {
data <- cbind(task="T1",
rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A1", value=0.8, case="C1")
))
expect_error(as.challenge(data, by="task", algorithm="algo", case="case", value="value", smallBetter=FALSE),
"The following case(s) appear(s) more than once for the same algorithm. Please revise. Or are you considering a multi-task challenge and forgot to specify argument 'by'?\nCase(s): C1", fixed=TRUE)
})
test_that("cases cannot appear more than once per algorithm for single-task challenge", {
data <- rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.7, case="C1"),
data.frame(algo="A1", value=0.5, case="C2"),
data.frame(algo="A2", value=0.6, case="C2"),
data.frame(algo="A2", value=0.6, case="C2"))
expect_error(as.challenge(data, taskName="T1", algorithm="algo", case="case", value="value", smallBetter=FALSE),
"The following case(s) appear(s) more than once for the same algorithm. Please revise. Or are you considering a multi-task challenge and forgot to specify argument 'by'?\nCase(s): C1, C2", fixed=TRUE)
})
test_that("cases cannot appear more than once per algorithm for multi-task challenge (1 task in data set)", {
data <- cbind(task="T1",
rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.7, case="C1"),
data.frame(algo="A1", value=0.5, case="C2"),
data.frame(algo="A2", value=0.6, case="C2"),
data.frame(algo="A2", value=0.6, case="C2")
))
expect_error(as.challenge(data, by="task", algorithm="algo", case="case", value="value", smallBetter=FALSE),
"The following case(s) appear(s) more than once for the same algorithm. Please revise. Or are you considering a multi-task challenge and forgot to specify argument 'by'?\nCase(s): C1, C2", fixed=TRUE)
})
test_that("cases cannot appear more than once per algorithm for multi-task challenge (2 tasks in data set)", {
dataTask1 <- cbind(task="T1",
rbind(
data.frame(algo="A1", value=0.8, case="C1") # let T1 pass
))
dataTask2 <- cbind(task="T2",
rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.7, case="C1"),
data.frame(algo="A1", value=0.5, case="C2"),
data.frame(algo="A2", value=0.6, case="C2"),
data.frame(algo="A2", value=0.6, case="C2")
))
data <- rbind(dataTask1, dataTask2)
expect_error(as.challenge(data, by="task", algorithm="algo", case="case", value="value", smallBetter=FALSE),
"The following case(s) appear(s) more than once for the same algorithm in task 'T2'. Please revise.\nCase(s): C1, C2", fixed=TRUE)
})
test_that("cases cannot appear more than once per algorithm when missing data was added for single-task challenge", {
data <- rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C2"),
data.frame(algo="A2", value=0.6, case="C2"))
expect_error(as.challenge(data, taskName="T1", algorithm="algo", case="case", value="value", smallBetter=FALSE),
"The following case(s) appear(s) more than once for the same algorithm. Please revise. Or are you considering a multi-task challenge and forgot to specify argument 'by'?\nCase(s): C1, C2", fixed=TRUE)
})
test_that("user is notified of duplicate cases when multi-task data set is interpreted as single-task data set (2 tasks in data set)", {
dataTask1 <- cbind(task="T1",
rbind(
data.frame(algo="A1", value=0.8, case="C1")
))
dataTask2 <- cbind(task="T2",
rbind(
data.frame(algo="A1", value=0.8, case="C1")
))
data <- rbind(dataTask1, dataTask2)
# do not specify parameter "by" to interpret multi-task data set as single-task data set
expect_error(as.challenge(data, taskName="New task", algorithm="algo", case="case", value="value", smallBetter=FALSE),
"The following case(s) appear(s) more than once for the same algorithm. Please revise. Or are you considering a multi-task challenge and forgot to specify argument 'by'?\nCase(s): C1", fixed=TRUE)
})
test_that("user is notified of missing algorithm performance when multi-task data set is interpreted as single-task data set (2 tasks in data set)", {
dataTask1 <- cbind(task="T1",
rbind(
data.frame(algo="A1", value=0.8, case="C1")
))
dataTask2 <- cbind(task="T2",
rbind(
data.frame(algo="A2", value=0.6, case="C2")
))
data <- rbind(dataTask1, dataTask2)
# do not specify parameter "by" to interpret multi-task data set as single-task data set
expect_message(as.challenge(data, taskName="New task", algorithm="algo", case="case", value="value", smallBetter=FALSE),
"Performance of not all algorithms has been observed for all cases. Therefore, missings have been inserted in the following cases:", fixed=TRUE)
})
test_that("NAs are replaced by numeric value for single-task challenge", {
data <- rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A1", value=NA, case="C2"),
data.frame(algo="A2", value=0.6, case="C1"),
data.frame(algo="A2", value=NA, case="C2"))
actualChallenge <- as.challenge(data, taskName="T1", algorithm="algo", case="case", value="value", smallBetter=FALSE, na.treat=0)
expect_equal(as.vector(actualChallenge$T1$algo), c("A1", "A1", "A2", "A2"))
expect_equal(as.vector(actualChallenge$T1$value), c(0.8, 0.0, 0.6, 0.0))
expect_equal(as.vector(actualChallenge$T1$case), c("C1", "C2", "C1", "C2"))
})
test_that("NAs are replaced by numeric value for multi-task challenge", {
dataTask1 <- cbind(task="T1",
rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A1", value=NA, case="C2")
))
dataTask2 <- cbind(task="T2",
rbind(
data.frame(algo="A2", value=NA, case="C1"),
data.frame(algo="A2", value=0.5, case="C2")
))
data <- rbind(dataTask1, dataTask2)
actualChallenge <- as.challenge(data, by="task", algorithm="algo", case="case", value="value", smallBetter=FALSE, na.treat=0)
expect_equal(as.vector(actualChallenge$T1$algo), c("A1", "A1"))
expect_equal(as.vector(actualChallenge$T1$value), c(0.8, 0.0))
expect_equal(as.vector(actualChallenge$T1$case), c("C1", "C2"))
expect_equal(as.vector(actualChallenge$T2$algo), c("A2", "A2"))
expect_equal(as.vector(actualChallenge$T2$value), c(0.0, 0.5))
expect_equal(as.vector(actualChallenge$T2$case), c("C1", "C2"))
})
test_that("NAs are replaced by function value for single-task challenge", {
data <- rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A1", value=NA, case="C2"),
data.frame(algo="A2", value=0.6, case="C1"),
data.frame(algo="A2", value=NA, case="C2"))
replacementFunction <- function(x) { 2 }
actualChallenge <- as.challenge(data, taskName="T1", algorithm="algo", case="case", value="value", smallBetter=FALSE, na.treat=replacementFunction)
expect_equal(as.vector(actualChallenge$T1$algo), c("A1", "A1", "A2", "A2"))
expect_equal(as.vector(actualChallenge$T1$value), c(0.8, 2.0, 0.6, 2.0))
expect_equal(as.vector(actualChallenge$T1$case), c("C1", "C2", "C1", "C2"))
})
test_that("NAs are replaced by function value for multi-task challenge", {
dataTask1 <- cbind(task="T1",
rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A1", value=NA, case="C2")
))
dataTask2 <- cbind(task="T2",
rbind(
data.frame(algo="A2", value=NA, case="C1"),
data.frame(algo="A2", value=0.5, case="C2")
))
data <- rbind(dataTask1, dataTask2)
replacementFunction <- function(x) { 2 }
actualChallenge <- as.challenge(data, by="task", algorithm="algo", case="case", value="value", smallBetter=FALSE, na.treat=replacementFunction)
expect_equal(as.vector(actualChallenge$T1$algo), c("A1", "A1"))
expect_equal(as.vector(actualChallenge$T1$value), c(0.8, 2.0))
expect_equal(as.vector(actualChallenge$T1$case), c("C1", "C2"))
expect_equal(as.vector(actualChallenge$T2$algo), c("A2", "A2"))
expect_equal(as.vector(actualChallenge$T2$value), c(2.0, 0.5))
expect_equal(as.vector(actualChallenge$T2$case), c("C1", "C2"))
})
test_that("NAs are removed for single-task challenge", {
data <- rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A1", value=NA, case="C2"),
data.frame(algo="A2", value=0.6, case="C1"),
data.frame(algo="A2", value=NA, case="C2"))
actualChallenge <- as.challenge(data, taskName="T1", algorithm="algo", case="case", value="value", smallBetter=FALSE, na.treat="na.rm")
expect_equal(as.vector(actualChallenge$T1$algo), c("A1", "A2"))
expect_equal(as.vector(actualChallenge$T1$value), c(0.8, 0.6))
expect_equal(as.vector(actualChallenge$T1$case), c("C1", "C1"))
})
test_that("NAs are removed for multi-task challenge", {
dataTask1 <- cbind(task="T1",
rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A1", value=NA, case="C2")
))
dataTask2 <- cbind(task="T2",
rbind(
data.frame(algo="A2", value=NA, case="C1"),
data.frame(algo="A2", value=0.5, case="C2")
))
data <- rbind(dataTask1, dataTask2)
actualChallenge <- as.challenge(data, by="task", algorithm="algo", case="case", value="value", smallBetter=FALSE, na.treat="na.rm")
expect_equal(as.vector(actualChallenge$T1$algo), c("A1"))
expect_equal(as.vector(actualChallenge$T1$value), c(0.8))
expect_equal(as.vector(actualChallenge$T1$case), c("C1"))
expect_equal(as.vector(actualChallenge$T2$algo), c("A2"))
expect_equal(as.vector(actualChallenge$T2$value), c(0.5))
expect_equal(as.vector(actualChallenge$T2$case), c("C2"))
})
test_that("automatically added NAs are replaced by numeric value for single-task challenge", {
data <- rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C2"))
expect_message(actualChallenge <- as.challenge(data, taskName="T1", algorithm="algo", case="case", value="value", smallBetter=FALSE, na.treat=0),
"Performance of not all algorithms has been observed for all cases. Therefore, missings have been inserted in the following cases:", fixed=TRUE)
expect_equal(as.vector(actualChallenge$T1$algo), c("A1", "A1", "A2", "A2"))
expect_equal(as.vector(actualChallenge$T1$value), c(0.8, 0.0, 0.0, 0.6))
expect_equal(as.vector(actualChallenge$T1$case), c("C1", "C2", "C1", "C2"))
})
test_that("automatically added NAs are replaced by numeric value for multi-task challenge", {
dataTask1 <- cbind(task="T1",
rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C2")
))
dataTask2 <- cbind(task="T2",
rbind(
data.frame(algo="A1", value=0.2, case="C1"),
data.frame(algo="A1", value=0.3, case="C2"),
data.frame(algo="A2", value=0.4, case="C1")
))
data <- rbind(dataTask1, dataTask2)
expect_message(actualChallenge <- as.challenge(data, by="task", algorithm="algo", case="case", value="value", smallBetter=FALSE, na.treat=0),
"Performance of not all algorithms has been observed for all cases in task 'T1'. Therefore, missings have been inserted in the following cases:", fixed=TRUE)
expect_equal(as.vector(actualChallenge$T1$algo), c("A1", "A1", "A2", "A2"))
expect_equal(as.vector(actualChallenge$T1$value), c(0.8, 0.0, 0.0, 0.6))
expect_equal(as.vector(actualChallenge$T1$case), c("C1", "C2", "C1", "C2"))
expect_equal(as.vector(actualChallenge$T2$algo), c("A1", "A1", "A2", "A2"))
expect_equal(as.vector(actualChallenge$T2$value), c(0.2, 0.3, 0.4, 0.0))
expect_equal(as.vector(actualChallenge$T2$case), c("C1", "C2", "C1", "C2"))
})
test_that("automatically added NAs are removed for single-task challenge", {
data <- rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C2"))
expect_message(actualChallenge <- as.challenge(data, taskName="T1", algorithm="algo", case="case", value="value", smallBetter=FALSE, na.treat="na.rm"),
"Performance of not all algorithms has been observed for all cases. Therefore, missings have been inserted in the following cases:", fixed=TRUE)
expect_equal(as.vector(actualChallenge$T1$algo), c("A1", "A2"))
expect_equal(as.vector(actualChallenge$T1$value), c(0.8, 0.6))
expect_equal(as.vector(actualChallenge$T1$case), c("C1", "C2"))
})
test_that("automatically added NAs are removed for multi-task challenge", {
dataTask1 <- cbind(task="T1",
rbind(
data.frame(algo="A1", value=0.8, case="C1"),
data.frame(algo="A2", value=0.6, case="C2")
))
dataTask2 <- cbind(task="T2",
rbind(
data.frame(algo="A1", value=0.2, case="C1"),
data.frame(algo="A1", value=0.3, case="C2"),
data.frame(algo="A2", value=0.4, case="C1")
))
data <- rbind(dataTask1, dataTask2)
expect_message(actualChallenge <- as.challenge(data, by="task", algorithm="algo", case="case", value="value", smallBetter=FALSE, na.treat="na.rm"),
"Performance of not all algorithms has been observed for all cases in task 'T1'. Therefore, missings have been inserted in the following cases:", fixed=TRUE)
expect_equal(as.vector(actualChallenge$T1$algo), c("A1", "A2"))
expect_equal(as.vector(actualChallenge$T1$value), c(0.8, 0.6))
expect_equal(as.vector(actualChallenge$T1$case), c("C1", "C2"))
expect_equal(as.vector(actualChallenge$T2$algo), c("A1", "A1", "A2"))
expect_equal(as.vector(actualChallenge$T2$value), c(0.2, 0.3, 0.4))
expect_equal(as.vector(actualChallenge$T2$case), c("C1", "C2", "C1"))
})
+test_that("class of 'algorithm' column must be 'factor' for single-task challenge", {
+ data <- rbind(
+ data.frame(algo="A1", value=0.8, case="C1"),
+ data.frame(algo="A1", value=0.7, case="C2"),
+ data.frame(algo="A2", value=0.6, case="C1"),
+ data.frame(algo="A2", value=0.5, case="C2"))
+
+ actualChallenge <- as.challenge(data, taskName="T1", algorithm="algo", case="case", value="value", smallBetter=FALSE, na.treat=0)
+
+ expect_equal(class(actualChallenge$T1$algo), "factor")
+})
+
+test_that("class of 'algorithm' column must be 'factor' for multi-task challenge", {
+ dataTask1 <- cbind(task="T1",
+ rbind(
+ data.frame(algo="A1", value=0.81, case="C1"),
+ data.frame(algo="A2", value=0.72, case="C1"),
+ data.frame(algo="A1", value=0.65, case="C2"),
+ data.frame(algo="A2", value=0.95, case="C2")
+ ))
+ dataTask2 <- cbind(task="T2",
+ rbind(
+ data.frame(algo="A1", value=0.75, case="C1"),
+ data.frame(algo="A2", value=0.82, case="C1"),
+ data.frame(algo="A1", value=0.66, case="C2"),
+ data.frame(algo="A2", value=0.84, case="C2")
+ ))
+
+ data <- rbind(dataTask1, dataTask2)
+
+ actualChallenge <- as.challenge(data, by="task", algorithm="algo", case="case", value="value", smallBetter=TRUE)
+
+ expect_equal(class(actualChallenge$T1$algo), "factor")
+ expect_equal(class(actualChallenge$T2$algo), "factor")
+})
+
+