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

Fix inplace predict missing value. (#6787)

Browse Source
This commit is contained in:
Jiaming Yuan
2021-03-27 05:36:10 +08:00
committed by GitHub
parent 5c87c2bba8
commit a59c7323b4
8 changed files with 97 additions and 33 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
tests/cpp/common/test_json.cc
View File

@@ -374,6 +374,11 @@ TEST(Json, AssigningNumber) {
value = 15; // NOLINT
ASSERT_EQ(get<Number>(json), 4);
}
{
Json value {Number(std::numeric_limits<float>::quiet_NaN())};
ASSERT_TRUE(IsA<Number>(value));
}
}
TEST(Json, AssigningString) {

27
tests/python-gpu/test_gpu_prediction.py
View File

@@ -154,17 +154,22 @@ class TestGPUPredict:
cp.cuda.runtime.setDevice(0)
rows = 1000
cols = 10
missing = 11 # set to integer for testing
cp_rng = cp.random.RandomState(1994)
cp.random.set_random_state(cp_rng)
X = cp.random.randn(rows, cols)
missing_idx = [i for i in range(0, cols, 4)]
X[:, missing_idx] = missing # set to be missing
y = cp.random.randn(rows)
dtrain = xgb.DMatrix(X, y)
booster = xgb.train({'tree_method': 'gpu_hist'},
dtrain, num_boost_round=10)
test = xgb.DMatrix(X[:10, ...])
predt_from_array = booster.inplace_predict(X[:10, ...])
booster = xgb.train({'tree_method': 'gpu_hist'}, dtrain, num_boost_round=10)
test = xgb.DMatrix(X[:10, ...], missing=missing)
predt_from_array = booster.inplace_predict(X[:10, ...], missing=missing)
predt_from_dmatrix = booster.predict(test)
cp.testing.assert_allclose(predt_from_array, predt_from_dmatrix)
@@ -185,6 +190,20 @@ class TestGPUPredict:
base_margin = cp_rng.randn(rows)
self.run_inplace_base_margin(booster, dtrain, X, base_margin)
# Create a wide dataset
X = cp_rng.randn(100, 10000)
y = cp_rng.randn(100)
missing_idx = [i for i in range(0, X.shape[1], 16)]
X[:, missing_idx] = missing
reg = xgb.XGBRegressor(tree_method="gpu_hist", n_estimators=8, missing=missing)
reg.fit(X, y)
gpu_predt = reg.predict(X)
reg.set_params(predictor="cpu_predictor")
cpu_predt = reg.predict(X)
np.testing.assert_allclose(gpu_predt, cpu_predt, atol=1e-6)
@pytest.mark.skipif(**tm.no_cudf())
def test_inplace_predict_cudf(self):
import cupy as cp

16
tests/python/test_predict.py
View File

@@ -103,31 +103,37 @@ class TestInplacePredict:
'''Tests for running inplace prediction'''
@classmethod
def setup_class(cls):
cls.rows = 100
cls.rows = 1000
cls.cols = 10
cls.missing = 11 # set to integer for testing
cls.rng = np.random.RandomState(1994)
cls.X = cls.rng.randn(cls.rows, cls.cols)
missing_idx = [i for i in range(0, cls.cols, 4)]
cls.X[:, missing_idx] = cls.missing # set to be missing
cls.y = cls.rng.randn(cls.rows)
dtrain = xgb.DMatrix(cls.X, cls.y)
cls.test = xgb.DMatrix(cls.X[:10, ...], missing=cls.missing)
cls.booster = xgb.train({'tree_method': 'hist'}, dtrain, num_boost_round=10)
cls.test = xgb.DMatrix(cls.X[:10, ...])
def test_predict(self):
booster = self.booster
X = self.X
test = self.test
predt_from_array = booster.inplace_predict(X[:10, ...])
predt_from_array = booster.inplace_predict(X[:10, ...], missing=self.missing)
predt_from_dmatrix = booster.predict(test)
np.testing.assert_allclose(predt_from_dmatrix, predt_from_array)
predt_from_array = booster.inplace_predict(X[:10, ...], iteration_range=(0, 4))
predt_from_array = booster.inplace_predict(
X[:10, ...], iteration_range=(0, 4), missing=self.missing
)
predt_from_dmatrix = booster.predict(test, ntree_limit=4)
np.testing.assert_allclose(predt_from_dmatrix, predt_from_array)