mirrors/cpython
1
0
Fork 0
mirror of https://github.com/python/cpython.git synced 2025-08-04 17:08:35 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 13

Minor speed/accuracy improvement for kde() (gh-119910)

Browse source
This commit is contained in:
Raymond Hettinger 2024-06-01 10:49:14 -05:00 • committed by GitHub
parent 90ec19fd33
commit ce2ea7d629
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 10 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

17
Lib/statistics.py
View file

@ -953,12 +953,14 @@ def kde(data, h, kernel='normal', *, cumulative=False):
case 'quartic' | 'biweight':
K = lambda t: 15/16 * (1.0 - t * t) ** 2
W = lambda t: 3/16 * t**5 - 5/8 * t**3 + 15/16 * t + 1/2
W = lambda t: sumprod((3/16, -5/8, 15/16, 1/2),
(t**5, t**3, t, 1.0))
support = 1.0
case 'triweight':
K = lambda t: 35/32 * (1.0 - t * t) ** 3
W = lambda t: 35/32 * (-1/7*t**7 + 3/5*t**5 - t**3 + t) + 1/2
W = lambda t: sumprod((-5/32, 21/32, -35/32, 35/32, 1/2),
(t**7, t**5, t**3, t, 1.0))
support = 1.0
case 'cosine':
@ -974,12 +976,10 @@ def kde(data, h, kernel='normal', *, cumulative=False):
if support is None:
def pdf(x):
n = len(data)
return sum(K((x - x_i) / h) for x_i in data) / (n * h)
return sum(K((x - x_i) / h) for x_i in data) / (len(data) * h)
def cdf(x):
n = len(data)
return sum(W((x - x_i) / h) for x_i in data) / n
return sum(W((x - x_i) / h) for x_i in data) / len(data)
else:
@ -1732,7 +1732,7 @@ def _quartic_invcdf_estimate(p):
_quartic_invcdf = _newton_raphson(
f_inv_estimate = _quartic_invcdf_estimate,
f = lambda t: 3/16 * t**5 - 5/8 * t**3 + 15/16 * t + 1/2,
f = lambda t: sumprod((3/16, -5/8, 15/16, 1/2), (t**5, t**3, t, 1.0)),
f_prime = lambda t: 15/16 * (1.0 - t * t) ** 2)
def _triweight_invcdf_estimate(p):
@ -1742,7 +1742,8 @@ def _triweight_invcdf_estimate(p):
_triweight_invcdf = _newton_raphson(
f_inv_estimate = _triweight_invcdf_estimate,
f = lambda t: 35/32 * (-1/7*t**7 + 3/5*t**5 - t**3 + t) + 1/2,
f = lambda t: sumprod((-5/32, 21/32, -35/32, 35/32, 1/2),
(t**7, t**5, t**3, t, 1.0)),
f_prime = lambda t: 35/32 * (1.0 - t * t) ** 3)
_kernel_invcdfs = {

2
Lib/test/test_statistics.py
View file

@ -2444,7 +2444,7 @@ class TestKDE(unittest.TestCase):
with self.subTest(kernel=kernel):
cdf = kde([0.0], h=1.0, kernel=kernel, cumulative=True)
for x in xarr:
self.assertAlmostEqual(invcdf(cdf(x)), x, places=5)
self.assertAlmostEqual(invcdf(cdf(x)), x, places=6)
@support.requires_resource('cpu')
def test_kde_random(self):