Hendrik/xgboost
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[Breaking] Don't drop trees during DART prediction by default (#5115)

Browse Source 
* Simplify DropTrees calling logic

* Add `training` parameter for prediction method.

* [Breaking]: Add `training` to C API.

* Change for R and Python custom objective.

* Correct comment.

Co-authored-by: Philip Hyunsu Cho <chohyu01@cs.washington.edu>
Co-authored-by: Jiaming Yuan <jm.yuan@outlook.com>
This commit is contained in:
Kodi Arfer
2020-01-13 08:48:30 -05:00
committed by Jiaming Yuan
parent 7b65698187
commit f100b8d878
23 changed files with 214 additions and 140 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
tests/python/test_basic_models.py
View File

@@ -44,7 +44,8 @@ class TestModels(unittest.TestCase):
def test_dart(self):
dtrain = xgb.DMatrix(dpath + 'agaricus.txt.train')
dtest = xgb.DMatrix(dpath + 'agaricus.txt.test')
param = {'max_depth': 5, 'objective': 'binary:logistic', 'booster': 'dart', 'verbosity': 1}
param = {'max_depth': 5, 'objective': 'binary:logistic',
'eval_metric': 'logloss', 'booster': 'dart', 'verbosity': 1}
# specify validations set to watch performance
watchlist = [(dtest, 'eval'), (dtrain, 'train')]
num_round = 2
@@ -52,7 +53,8 @@ class TestModels(unittest.TestCase):
# this is prediction
preds = bst.predict(dtest, ntree_limit=num_round)
labels = dtest.get_label()
err = sum(1 for i in range(len(preds)) if int(preds[i] > 0.5) != labels[i]) / float(len(preds))
err = sum(1 for i in range(len(preds))
if int(preds[i] > 0.5) != labels[i]) / float(len(preds))
# error must be smaller than 10%
assert err < 0.1
@@ -68,18 +70,31 @@ class TestModels(unittest.TestCase):
# assert they are the same
assert np.sum(np.abs(preds2 - preds)) == 0
def my_logloss(preds, dtrain):
labels = dtrain.get_label()
return 'logloss', np.sum(
np.log(np.where(labels, preds, 1 - preds)))
# check whether custom evaluation metrics work
bst = xgb.train(param, dtrain, num_round, watchlist,
feval=my_logloss)
preds3 = bst.predict(dtest, ntree_limit=num_round)
assert all(preds3 == preds)
# check whether sample_type and normalize_type work
num_round = 50
param['verbosity'] = 0
param['learning_rate'] = 0.1
param['rate_drop'] = 0.1
preds_list = []
for p in [[p0, p1] for p0 in ['uniform', 'weighted'] for p1 in ['tree', 'forest']]:
for p in [[p0, p1] for p0 in ['uniform', 'weighted']
for p1 in ['tree', 'forest']]:
param['sample_type'] = p[0]
param['normalize_type'] = p[1]
bst = xgb.train(param, dtrain, num_round, watchlist)
preds = bst.predict(dtest, ntree_limit=num_round)
err = sum(1 for i in range(len(preds)) if int(preds[i] > 0.5) != labels[i]) / float(len(preds))
err = sum(1 for i in range(len(preds))
if int(preds[i] > 0.5) != labels[i]) / float(len(preds))
assert err < 0.1
preds_list.append(preds)