2016-06-09 02:53:37 -05:00

160 lines
6.0 KiB
R

require(xgboost)
context("basic functions")
data(agaricus.train, package='xgboost')
data(agaricus.test, package='xgboost')
train <- agaricus.train
test <- agaricus.test
set.seed(1994)
test_that("train and predict binary classification", {
nround = 2
expect_output(
bst <- xgboost(data = train$data, label = train$label, max.depth = 2,
eta = 1, nthread = 2, nround = nround, objective = "binary:logistic")
, "train-error")
expect_equal(class(bst), "xgb.Booster")
expect_true(!is.null(bst$evaluation_log))
expect_equal(nrow(bst$evaluation_log), nround)
expect_lt(bst$evaluation_log[, min(train_error)], 0.03)
expect_equal(bst$nboost, bst$ntree)
pred <- predict(bst, test$data)
expect_equal(length(pred), 1611)
})
test_that("train and predict softprob", {
expect_output(
bst <- xgboost(data = as.matrix(iris[, -5]), label = as.numeric(iris$Species) - 1,
max.depth = 3, eta = 0.5, nthread = 2, nround = 5,
objective = "multi:softprob", num_class=3)
, "train-merror")
expect_true(!is.null(bst$evaluation_log))
expect_lt(bst$evaluation_log[, min(train_merror)], 0.025)
expect_equal(bst$nboost * 3, bst$ntree)
pred <- predict(bst, as.matrix(iris[, -5]))
expect_equal(length(pred), nrow(iris) * 3)
})
test_that("train and predict softmax", {
expect_output(
bst <- xgboost(data = as.matrix(iris[, -5]), label = as.numeric(iris$Species) - 1,
max.depth = 3, eta = 0.15, nthread = 2, nround = 25,
objective = "multi:softmax", num_class=3)
, "train-merror")
expect_true(!is.null(bst$evaluation_log))
expect_lt(bst$evaluation_log[, min(train_merror)], 0.025)
expect_equal(bst$nboost * 3, bst$ntree)
pred <- predict(bst, as.matrix(iris[, -5]))
expect_equal(length(pred), nrow(iris))
})
test_that("early stopping xgb.train works", {
expect_output(
bst <- xgboost(data = train$data, label = train$label, max.depth = 2,
eta = 0.3, nthread = 2, nround = 20, objective = "binary:logistic",
early.stop.round = 3, maximize = FALSE)
, "Stopping. Best iteration")
expect_true(!is.null(bst$best_iteration))
expect_lt(bst$best_iteration, 19)
expect_equal(bst$nboost, bst$ntree)
expect_equal(bst$best_iteration, bst$best_ntreelimit)
pred <- predict(bst, test$data)
expect_equal(length(pred), 1611)
})
test_that("training continuation works", {
dtrain <- xgb.DMatrix(train$data, label = train$label)
watchlist = list(train=dtrain)
param <- list(objective = "binary:logistic", max.depth = 2, eta = 1, nthread = 2)
# for the reference, use 4 iterations at once:
set.seed(11)
bst <- xgb.train(param, dtrain, nrounds = 4, watchlist)
# first two iterations:
set.seed(11)
bst1 <- xgb.train(param, dtrain, nrounds = 2, watchlist)
# continue for two more:
bst2 <- xgb.train(param, dtrain, nrounds = 2, watchlist, xgb_model = bst1)
expect_equal(bst$raw, bst2$raw)
expect_true(!is.null(bst2$evaluation_log))
expect_equal(dim(bst2$evaluation_log), c(4, 2))
expect_equal(bst2$evaluation_log, bst$evaluation_log)
# test continuing from raw model data
bst2 <- xgb.train(param, dtrain, nrounds = 2, watchlist, xgb_model = bst1$raw)
expect_equal(bst$raw, bst2$raw)
expect_equal(dim(bst2$evaluation_log), c(2, 2))
# test continuing from a model in file
xgb.save(bst1, "xgboost.model")
bst2 <- xgb.train(param, dtrain, nrounds = 2, watchlist, xgb_model = "xgboost.model")
expect_equal(bst$raw, bst2$raw)
expect_equal(dim(bst2$evaluation_log), c(2, 2))
})
test_that("xgb.cv works", {
cv <- xgb.cv(data = train$data, label = train$label, max.depth = 2, nfold = 5,
eta = 1., nthread = 2, nround = 2, objective = "binary:logistic",
verbose=TRUE)
expect_is(cv, 'xgb.cv.synchronous')
expect_true(!is.null(cv$evaluation_log))
expect_lt(cv$evaluation_log[, min(test_error_mean)], 0.03)
expect_lt(cv$evaluation_log[, min(test_error_std)], 0.004)
expect_equal(cv$nboost, cv$ntree)
expect_true(!is.null(cv$folds) && is.list(cv$folds))
expect_length(cv$folds, 5)
expect_true(!is.null(cv$params) && is.list(cv$params))
expect_true(!is.null(cv$callbacks))
expect_true(!is.null(cv$call))
})
test_that("early stopping xgb.cv works", {
expect_output(
cv <- xgb.cv(data = train$data, label = train$label, max.depth = 2, nfold = 5,
eta = 0.5, nthread = 2, nround = 20, objective = "binary:logistic",
early.stop.round = 3, maximize = FALSE, verbose=TRUE)
, "Stopping. Best iteration")
expect_true(!is.null(cv$best_iteration))
expect_lt(cv$best_iteration, 19)
expect_equal(cv$nboost, cv$ntree)
expect_equal(cv$best_iteration, cv$best_ntreelimit)
})
test_that("prediction in xgb.cv works", {
nround = 2
cv <- xgb.cv(data = train$data, label = train$label, max.depth = 2, nfold = 5,
eta = 0.5, nthread = 2, nround = nround, objective = "binary:logistic",
verbose=TRUE, prediction = T)
expect_true(!is.null(cv$evaluation_log))
expect_true(!is.null(cv$pred))
expect_length(cv$pred, nrow(train$data))
err_pred <- sum((cv$pred > 0.5) != train$label)/length(train$label)
err_log <- cv$evaluation_log[nround, test_error_mean]
expect_lt(abs(err_pred - err_log), 10e-6)
})
test_that("prediction in early-stopping xgb.cv works", {
set.seed(123)
# add some label noise
lb <- train$label
lb[sample(length(train$label), 2000)] <- 0
expect_output(
cv <- xgb.cv(data = train$data, label = lb, max.depth = 3, nfold = 5,
eta = 1., nthread = 2, nround = 20, objective = "binary:logistic",
early.stop.round = 3, maximize = FALSE, verbose=TRUE, predict=TRUE)
, "Stopping. Best iteration")
expect_true(!is.null(cv$best_iteration))
expect_lt(cv$best_iteration, 19)
expect_true(!is.null(cv$evaluation_log))
expect_true(!is.null(cv$pred))
expect_length(cv$pred, nrow(train$data))
err_pred <- sum((cv$pred > 0.5) != lb)/length(lb)
err_log <- cv$evaluation_log[cv$best_iteration, test_error_mean]
expect_lt(abs(err_pred - err_log), 10e-6)
})