1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
|
import numpy as np
import pytest
from numpy.testing import assert_almost_equal, assert_array_equal, assert_equal
from scipy import stats
from sklearn.ensemble import GradientBoostingRegressor, RandomForestRegressor
from skopt.learning import GradientBoostingQuantileRegressor
def truth(X):
return 0.5 * np.sin(1.75 * X[:, 0])
@pytest.mark.fast_test
def test_gbrt_gaussian():
# estimate quantiles of the normal distribution
rng = np.random.RandomState(1)
N = 10000
X = np.ones((N, 1))
y = rng.normal(size=N)
rgr = GradientBoostingQuantileRegressor()
rgr.fit(X, y)
estimates = rgr.predict(X, return_quantiles=True)
assert_almost_equal(
stats.norm.ppf(rgr.quantiles), np.mean(estimates, axis=0), decimal=2
)
@pytest.mark.fast_test
def test_gbrt_base_estimator():
rng = np.random.RandomState(1)
N = 10000
X = np.ones((N, 1))
y = rng.normal(size=N)
base = RandomForestRegressor()
rgr = GradientBoostingQuantileRegressor(base_estimator=base)
with pytest.raises(ValueError):
# 'type GradientBoostingRegressor',
rgr.fit(X, y)
base = GradientBoostingRegressor()
rgr = GradientBoostingQuantileRegressor(base_estimator=base)
with pytest.raises(ValueError):
# 'quantile loss'
rgr.fit(X, y)
base = GradientBoostingRegressor(loss='quantile', n_estimators=20)
rgr = GradientBoostingQuantileRegressor(base_estimator=base)
rgr.fit(X, y)
estimates = rgr.predict(X, return_quantiles=True)
assert_almost_equal(
stats.norm.ppf(rgr.quantiles), np.mean(estimates, axis=0), decimal=2
)
@pytest.mark.fast_test
def test_gbrt_with_std():
# simple test of the interface
rng = np.random.RandomState(1)
X = rng.uniform(0, 5, 500)[:, np.newaxis]
noise_level = 0.5
y = truth(X) + rng.normal(0, noise_level, len(X))
X_ = np.linspace(0, 5, 1000)[:, np.newaxis]
model = GradientBoostingQuantileRegressor()
model.fit(X, y)
# three quantiles, so three numbers per sample
assert_array_equal(model.predict(X_, return_quantiles=True).shape, (len(X_), 3))
# "traditional" API which returns one number per sample, in this case
# just the median/mean
assert_array_equal(model.predict(X_).shape, (len(X_)))
l, c, h = model.predict(X_, return_quantiles=True).T
assert_equal(l.shape, c.shape)
assert_equal(c.shape, h.shape)
assert_equal(l.shape[0], X_.shape[0])
mean, std = model.predict(X_, return_std=True)
assert_array_equal(mean, c)
assert_array_equal(std, (h - l) / 2.0)
@pytest.mark.fast_test
def test_gbrt_in_parallel():
# check estimate quantiles with parallel
rng = np.random.RandomState(1)
N = 10000
X = np.ones((N, 1))
y = rng.normal(size=N)
rgr = GradientBoostingQuantileRegressor(
n_jobs=1, random_state=np.random.RandomState(1)
)
rgr.fit(X, y)
estimates = rgr.predict(X)
rgr.set_params(n_jobs=2, random_state=np.random.RandomState(1))
rgr.fit(X, y)
estimates_parallel = rgr.predict(X)
assert_array_equal(estimates, estimates_parallel)
|
