xgboost/tests/python/test_updaters.py
Jiaming Yuan fdf533f2b9
[POC] Experimental support for l1 error. (#7812)
Support adaptive tree, a feature supported by both sklearn and lightgbm.  The tree leaf is recomputed based on residue of labels and predictions after construction.

For l1 error, the optimal value is the median (50 percentile).

This is marked as experimental support for the following reasons:
- The value is not well defined for distributed training, where we might have empty leaves for local workers. Right now I just use the original leaf value for computing the average with other workers, which might cause significant errors.
- Some follow-ups are required, for exact, pruner, and optimization for quantile function. Also, we need to calculate the initial estimation.
2022-04-26 21:41:55 +08:00

251 lines
9.6 KiB
Python

import testing as tm
import pytest
import xgboost as xgb
import numpy as np
from hypothesis import given, strategies, settings, note
exact_parameter_strategy = strategies.fixed_dictionaries({
'nthread': strategies.integers(1, 4),
'max_depth': strategies.integers(1, 11),
'min_child_weight': strategies.floats(0.5, 2.0),
'alpha': strategies.floats(0.0, 2.0),
'lambda': strategies.floats(1e-5, 2.0),
'eta': strategies.floats(0.01, 0.5),
'gamma': strategies.floats(0.0, 2.0),
'seed': strategies.integers(0, 10),
# We cannot enable subsampling as the training loss can increase
# 'subsample': strategies.floats(0.5, 1.0),
'colsample_bytree': strategies.floats(0.5, 1.0),
'colsample_bylevel': strategies.floats(0.5, 1.0),
})
hist_parameter_strategy = strategies.fixed_dictionaries({
'max_depth': strategies.integers(1, 11),
'max_leaves': strategies.integers(0, 1024),
'max_bin': strategies.integers(2, 512),
'grow_policy': strategies.sampled_from(['lossguide', 'depthwise']),
}).filter(lambda x: (x['max_depth'] > 0 or x['max_leaves'] > 0) and (
x['max_depth'] > 0 or x['grow_policy'] == 'lossguide'))
def train_result(param, dmat, num_rounds):
result = {}
xgb.train(param, dmat, num_rounds, [(dmat, 'train')], verbose_eval=False,
evals_result=result)
return result
class TestTreeMethod:
@given(exact_parameter_strategy, strategies.integers(1, 20),
tm.dataset_strategy)
@settings(deadline=None, print_blob=True)
def test_exact(self, param, num_rounds, dataset):
if dataset.name.endswith("-l1"):
return
param['tree_method'] = 'exact'
param = dataset.set_params(param)
result = train_result(param, dataset.get_dmat(), num_rounds)
assert tm.non_increasing(result['train'][dataset.metric])
@given(
exact_parameter_strategy,
hist_parameter_strategy,
strategies.integers(1, 20),
tm.dataset_strategy,
)
@settings(deadline=None, print_blob=True)
def test_approx(self, param, hist_param, num_rounds, dataset):
param["tree_method"] = "approx"
param = dataset.set_params(param)
param.update(hist_param)
result = train_result(param, dataset.get_dmat(), num_rounds)
note(result)
assert tm.non_increasing(result["train"][dataset.metric])
@pytest.mark.skipif(**tm.no_sklearn())
def test_pruner(self):
import sklearn
params = {'tree_method': 'exact'}
cancer = sklearn.datasets.load_breast_cancer()
X = cancer['data']
y = cancer["target"]
dtrain = xgb.DMatrix(X, y)
booster = xgb.train(params, dtrain=dtrain, num_boost_round=10)
grown = str(booster.get_dump())
params = {'updater': 'prune', 'process_type': 'update', 'gamma': '0.2'}
booster = xgb.train(params, dtrain=dtrain, num_boost_round=10,
xgb_model=booster)
after_prune = str(booster.get_dump())
assert grown != after_prune
booster = xgb.train(params, dtrain=dtrain, num_boost_round=10,
xgb_model=booster)
second_prune = str(booster.get_dump())
# Second prune should not change the tree
assert after_prune == second_prune
@given(exact_parameter_strategy, hist_parameter_strategy, strategies.integers(1, 20),
tm.dataset_strategy)
@settings(deadline=None, print_blob=True)
def test_hist(self, param, hist_param, num_rounds, dataset):
param['tree_method'] = 'hist'
param = dataset.set_params(param)
param.update(hist_param)
result = train_result(param, dataset.get_dmat(), num_rounds)
note(result)
assert tm.non_increasing(result['train'][dataset.metric])
def test_hist_categorical(self):
# hist must be same as exact on all-categorial data
dpath = 'demo/data/'
ag_dtrain = xgb.DMatrix(dpath + 'agaricus.txt.train')
ag_dtest = xgb.DMatrix(dpath + 'agaricus.txt.test')
ag_param = {'max_depth': 2,
'tree_method': 'hist',
'eta': 1,
'verbosity': 0,
'objective': 'binary:logistic',
'eval_metric': 'auc'}
hist_res = {}
exact_res = {}
xgb.train(ag_param, ag_dtrain, 10,
[(ag_dtrain, 'train'), (ag_dtest, 'test')],
evals_result=hist_res)
ag_param["tree_method"] = "exact"
xgb.train(ag_param, ag_dtrain, 10,
[(ag_dtrain, 'train'), (ag_dtest, 'test')],
evals_result=exact_res)
assert hist_res['train']['auc'] == exact_res['train']['auc']
assert hist_res['test']['auc'] == exact_res['test']['auc']
@pytest.mark.skipif(**tm.no_sklearn())
def test_hist_degenerate_case(self):
# Test a degenerate case where the quantile sketcher won't return any
# quantile points for a particular feature (the second feature in
# this example). Source: https://github.com/dmlc/xgboost/issues/2943
nan = np.nan
param = {'missing': nan, 'tree_method': 'hist'}
model = xgb.XGBRegressor(**param)
X = np.array([[6.18827160e+05, 1.73000000e+02], [6.37345679e+05, nan],
[6.38888889e+05, nan], [6.28086420e+05, nan]])
y = [1000000., 0., 0., 500000.]
w = [0, 0, 1, 0]
model.fit(X, y, sample_weight=w)
def run_invalid_category(self, tree_method: str) -> None:
rng = np.random.default_rng()
# too large
X = rng.integers(low=0, high=4, size=1000).reshape(100, 10)
y = rng.normal(loc=0, scale=1, size=100)
X[13, 7] = np.iinfo(np.int32).max + 1
# Check is performed during sketching.
Xy = xgb.DMatrix(X, y, feature_types=["c"] * 10)
with pytest.raises(ValueError):
xgb.train({"tree_method": tree_method}, Xy)
X[13, 7] = 16777216
Xy = xgb.DMatrix(X, y, feature_types=["c"] * 10)
with pytest.raises(ValueError):
xgb.train({"tree_method": tree_method}, Xy)
# mixed positive and negative values
X = rng.normal(loc=0, scale=1, size=1000).reshape(100, 10)
y = rng.normal(loc=0, scale=1, size=100)
Xy = xgb.DMatrix(X, y, feature_types=["c"] * 10)
with pytest.raises(ValueError):
xgb.train({"tree_method": tree_method}, Xy)
if tree_method == "gpu_hist":
import cupy as cp
X, y = cp.array(X), cp.array(y)
with pytest.raises(ValueError):
Xy = xgb.DeviceQuantileDMatrix(X, y, feature_types=["c"] * 10)
def test_invalid_category(self) -> None:
self.run_invalid_category("approx")
def run_categorical_basic(self, rows, cols, rounds, cats, tree_method):
onehot, label = tm.make_categorical(rows, cols, cats, True)
cat, _ = tm.make_categorical(rows, cols, cats, False)
by_etl_results = {}
by_builtin_results = {}
predictor = "gpu_predictor" if tree_method == "gpu_hist" else None
# Use one-hot exclusively
parameters = {
"tree_method": tree_method, "predictor": predictor, "max_cat_to_onehot": 9999
}
m = xgb.DMatrix(onehot, label, enable_categorical=False)
xgb.train(
parameters,
m,
num_boost_round=rounds,
evals=[(m, "Train")],
evals_result=by_etl_results,
)
m = xgb.DMatrix(cat, label, enable_categorical=True)
xgb.train(
parameters,
m,
num_boost_round=rounds,
evals=[(m, "Train")],
evals_result=by_builtin_results,
)
# There are guidelines on how to specify tolerance based on considering output as
# random variables. But in here the tree construction is extremely sensitive to
# floating point errors. An 1e-5 error in a histogram bin can lead to an entirely
# different tree. So even though the test is quite lenient, hypothesis can still
# pick up falsifying examples from time to time.
np.testing.assert_allclose(
np.array(by_etl_results["Train"]["rmse"]),
np.array(by_builtin_results["Train"]["rmse"]),
rtol=1e-3,
)
assert tm.non_increasing(by_builtin_results["Train"]["rmse"])
by_grouping: xgb.callback.TrainingCallback.EvalsLog = {}
parameters["max_cat_to_onehot"] = 1
parameters["reg_lambda"] = 0
m = xgb.DMatrix(cat, label, enable_categorical=True)
xgb.train(
parameters,
m,
num_boost_round=rounds,
evals=[(m, "Train")],
evals_result=by_grouping,
)
rmse_oh = by_builtin_results["Train"]["rmse"]
rmse_group = by_grouping["Train"]["rmse"]
# always better or equal to onehot when there's no regularization.
for a, b in zip(rmse_oh, rmse_group):
assert a >= b
parameters["reg_lambda"] = 1.0
by_grouping = {}
xgb.train(
parameters,
m,
num_boost_round=32,
evals=[(m, "Train")],
evals_result=by_grouping,
)
assert tm.non_increasing(by_grouping["Train"]["rmse"]), by_grouping
@given(strategies.integers(10, 400), strategies.integers(3, 8),
strategies.integers(1, 2), strategies.integers(4, 7))
@settings(deadline=None, print_blob=True)
@pytest.mark.skipif(**tm.no_pandas())
def test_categorical(self, rows, cols, rounds, cats):
self.run_categorical_basic(rows, cols, rounds, cats, "approx")
self.run_categorical_basic(rows, cols, rounds, cats, "hist")