From 11c79531655a4aa3f82c20ff562ac571f40040cc Mon Sep 17 00:00:00 2001 From: Raymond Hettinger Date: Sat, 23 Feb 2019 14:44:07 -0800 Subject: [PATCH] bpo-36018: Add the NormalDist class to the statistics module (GH-11973) --- Doc/library/statistics.rst | 195 ++++++++++++++++++ Doc/whatsnew/3.8.rst | 26 +++ Lib/statistics.py | 156 +++++++++++++- Lib/test/test_statistics.py | 177 ++++++++++++++++ .../2019-02-21-15-47-00.bpo-36018.qt7QUe.rst | 3 + 5 files changed, 556 insertions(+), 1 deletion(-) create mode 100644 Misc/NEWS.d/next/Library/2019-02-21-15-47-00.bpo-36018.qt7QUe.rst diff --git a/Doc/library/statistics.rst b/Doc/library/statistics.rst index 20a2c1cb13e..c1be295cbf9 100644 --- a/Doc/library/statistics.rst +++ b/Doc/library/statistics.rst @@ -467,6 +467,201 @@ A single exception is defined: Subclass of :exc:`ValueError` for statistics-related exceptions. + +:class:`NormalDist` objects +=========================== + +A :class:`NormalDist` is a a composite class that treats the mean and standard +deviation of data measurements as a single entity. It is a tool for creating +and manipulating normal distributions of a random variable. + +Normal distributions arise from the `Central Limit Theorem +`_ and have a wide range +of applications in statistics, including simulations and hypothesis testing. + +.. class:: NormalDist(mu=0.0, sigma=1.0) + + Returns a new *NormalDist* object where *mu* represents the `arithmetic + mean `_ of data and *sigma* + represents the `standard deviation + `_ of the data. + + If *sigma* is negative, raises :exc:`StatisticsError`. + + .. attribute:: mu + + The mean of a normal distribution. + + .. attribute:: sigma + + The standard deviation of a normal distribution. + + .. attribute:: variance + + A read-only property representing the `variance + `_ of a normal + distribution. Equal to the square of the standard deviation. + + .. classmethod:: NormalDist.from_samples(data) + + Class method that makes a normal distribution instance + from sample data. The *data* can be any :term:`iterable` + and should consist of values that can be converted to type + :class:`float`. + + If *data* does not contain at least two elements, raises + :exc:`StatisticsError` because it takes at least one point to estimate + a central value and at least two points to estimate dispersion. + + .. method:: NormalDist.samples(n, seed=None) + + Generates *n* random samples for a given mean and standard deviation. + Returns a :class:`list` of :class:`float` values. + + If *seed* is given, creates a new instance of the underlying random + number generator. This is useful for creating reproducible results, + even in a multi-threading context. + + .. method:: NormalDist.pdf(x) + + Using a `probability density function (pdf) + `_, + compute the relative likelihood that a random sample *X* will be near + the given value *x*. Mathematically, it is the ratio ``P(x <= X < + x+dx) / dx``. + + Note the relative likelihood of *x* can be greater than `1.0`. The + probability for a specific point on a continuous distribution is `0.0`, + so the :func:`pdf` is used instead. It gives the probability of a + sample occurring in a narrow range around *x* and then dividing that + probability by the width of the range (hence the word "density"). + + .. method:: NormalDist.cdf(x) + + Using a `cumulative distribution function (cdf) + `_, + compute the probability that a random sample *X* will be less than or + equal to *x*. Mathematically, it is written ``P(X <= x)``. + + Instances of :class:`NormalDist` support addition, subtraction, + multiplication and division by a constant. These operations + are used for translation and scaling. For example: + + .. doctest:: + + >>> temperature_february = NormalDist(5, 2.5) # Celsius + >>> temperature_february * (9/5) + 32 # Fahrenheit + NormalDist(mu=41.0, sigma=4.5) + + Dividing a constant by an instance of :class:`NormalDist` is not supported. + + Since normal distributions arise from additive effects of independent + variables, it is possible to `add and subtract two normally distributed + random variables + `_ + represented as instances of :class:`NormalDist`. For example: + + .. doctest:: + + >>> birth_weights = NormalDist.from_samples([2.5, 3.1, 2.1, 2.4, 2.7, 3.5]) + >>> drug_effects = NormalDist(0.4, 0.15) + >>> combined = birth_weights + drug_effects + >>> f'mu={combined.mu :.1f} sigma={combined.sigma :.1f}' + 'mu=3.1 sigma=0.5' + + .. versionadded:: 3.8 + + +:class:`NormalDist` Examples and Recipes +---------------------------------------- + +A :class:`NormalDist` readily solves classic probability problems. + +For example, given `historical data for SAT exams +`_ showing that scores +are normally distributed with a mean of 1060 and standard deviation of 192, +determine the percentage of students with scores between 1100 and 1200: + +.. doctest:: + + >>> sat = NormalDist(1060, 195) + >>> fraction = sat.cdf(1200) - sat.cdf(1100) + >>> f'{fraction * 100 :.1f}% score between 1100 and 1200' + '18.2% score between 1100 and 1200' + +To estimate the distribution for a model than isn't easy to solve +analytically, :class:`NormalDist` can generate input samples for a `Monte +Carlo simulation `_ of the +model: + +.. doctest:: + + >>> n = 100_000 + >>> X = NormalDist(350, 15).samples(n) + >>> Y = NormalDist(47, 17).samples(n) + >>> Z = NormalDist(62, 6).samples(n) + >>> model_simulation = [x * y / z for x, y, z in zip(X, Y, Z)] + >>> NormalDist.from_samples(model_simulation) # doctest: +SKIP + NormalDist(mu=267.6516398754636, sigma=101.357284306067) + +Normal distributions commonly arise in machine learning problems. + +Wikipedia has a `nice example with a Naive Bayesian Classifier +`_. The challenge +is to guess a person's gender from measurements of normally distributed +features including height, weight, and foot size. + +The `prior probability `_ of +being male or female is 50%: + +.. doctest:: + + >>> prior_male = 0.5 + >>> prior_female = 0.5 + +We also have a training dataset with measurements for eight people. These +measurements are assumed to be normally distributed, so we summarize the data +with :class:`NormalDist`: + +.. doctest:: + + >>> height_male = NormalDist.from_samples([6, 5.92, 5.58, 5.92]) + >>> height_female = NormalDist.from_samples([5, 5.5, 5.42, 5.75]) + >>> weight_male = NormalDist.from_samples([180, 190, 170, 165]) + >>> weight_female = NormalDist.from_samples([100, 150, 130, 150]) + >>> foot_size_male = NormalDist.from_samples([12, 11, 12, 10]) + >>> foot_size_female = NormalDist.from_samples([6, 8, 7, 9]) + +We observe a new person whose feature measurements are known but whose gender +is unknown: + +.. doctest:: + + >>> ht = 6.0 # height + >>> wt = 130 # weight + >>> fs = 8 # foot size + +The posterior is the product of the prior times each likelihood of a +feature measurement given the gender: + +.. doctest:: + + >>> posterior_male = (prior_male * height_male.pdf(ht) * + ... weight_male.pdf(wt) * foot_size_male.pdf(fs)) + + >>> posterior_female = (prior_female * height_female.pdf(ht) * + ... weight_female.pdf(wt) * foot_size_female.pdf(fs)) + +The final prediction is awarded to the largest posterior -- this is known as +the `maximum a posteriori +`_ or MAP: + +.. doctest:: + + >>> 'male' if posterior_male > posterior_female else 'female' + 'female' + + .. # This modelines must appear within the last ten lines of the file. kate: indent-width 3; remove-trailing-space on; replace-tabs on; encoding utf-8; diff --git a/Doc/whatsnew/3.8.rst b/Doc/whatsnew/3.8.rst index f21175a5a2a..68a4457106b 100644 --- a/Doc/whatsnew/3.8.rst +++ b/Doc/whatsnew/3.8.rst @@ -278,6 +278,32 @@ Added :func:`statistics.fmean` as a faster, floating point variant of :func:`statistics.mean()`. (Contributed by Raymond Hettinger and Steven D'Aprano in :issue:`35904`.) +Added :class:`statistics.NormalDist`, a tool for creating +and manipulating normal distributions of a random variable. +(Contributed by Raymond Hettinger in :issue:`36018`.) + +:: + + >>> temperature_feb = NormalDist.from_samples([4, 12, -3, 2, 7, 14]) + >>> temperature_feb + NormalDist(mu=6.0, sigma=6.356099432828281) + + >>> temperature_feb.cdf(3) # Chance of being under 3 degrees + 0.3184678262814532 + >>> # Relative chance of being 7 degrees versus 10 degrees + >>> temperature_feb.pdf(7) / temperature_feb.pdf(10) + 1.2039930378537762 + + >>> el_nino = NormalDist(4, 2.5) + >>> temperature_feb += el_nino # Add in a climate effect + >>> temperature_feb + NormalDist(mu=10.0, sigma=6.830080526611674) + + >>> temperature_feb * (9/5) + 32 # Convert to Fahrenheit + NormalDist(mu=50.0, sigma=12.294144947901014) + >>> temperature_feb.samples(3) # Generate random samples + [7.672102882379219, 12.000027119750287, 4.647488369766392] + tokenize -------- diff --git a/Lib/statistics.py b/Lib/statistics.py index 8ecb906d869..a73001ac554 100644 --- a/Lib/statistics.py +++ b/Lib/statistics.py @@ -76,7 +76,7 @@ A single exception is defined: StatisticsError is a subclass of ValueError. """ -__all__ = [ 'StatisticsError', +__all__ = [ 'StatisticsError', 'NormalDist', 'pstdev', 'pvariance', 'stdev', 'variance', 'median', 'median_low', 'median_high', 'median_grouped', 'mean', 'mode', 'harmonic_mean', 'fmean', @@ -85,11 +85,13 @@ __all__ = [ 'StatisticsError', import collections import math import numbers +import random from fractions import Fraction from decimal import Decimal from itertools import groupby from bisect import bisect_left, bisect_right +from math import hypot, sqrt, fabs, exp, erf, tau @@ -694,3 +696,155 @@ def pstdev(data, mu=None): return var.sqrt() except AttributeError: return math.sqrt(var) + +## Normal Distribution ##################################################### + +class NormalDist: + 'Normal distribution of a random variable' + # https://en.wikipedia.org/wiki/Normal_distribution + # https://en.wikipedia.org/wiki/Variance#Properties + + __slots__ = ('mu', 'sigma') + + def __init__(self, mu=0.0, sigma=1.0): + 'NormalDist where mu is the mean and sigma is the standard deviation' + if sigma < 0.0: + raise StatisticsError('sigma must be non-negative') + self.mu = mu + self.sigma = sigma + + @classmethod + def from_samples(cls, data): + 'Make a normal distribution instance from sample data' + if not isinstance(data, (list, tuple)): + data = list(data) + xbar = fmean(data) + return cls(xbar, stdev(data, xbar)) + + def samples(self, n, seed=None): + 'Generate *n* samples for a given mean and standard deviation' + gauss = random.gauss if seed is None else random.Random(seed).gauss + mu, sigma = self.mu, self.sigma + return [gauss(mu, sigma) for i in range(n)] + + def pdf(self, x): + 'Probability density function: P(x <= X < x+dx) / dx' + variance = self.sigma ** 2.0 + if not variance: + raise StatisticsError('pdf() not defined when sigma is zero') + return exp((x - self.mu)**2.0 / (-2.0*variance)) / sqrt(tau * variance) + + def cdf(self, x): + 'Cumulative density function: P(X <= x)' + if not self.sigma: + raise StatisticsError('cdf() not defined when sigma is zero') + return 0.5 * (1.0 + erf((x - self.mu) / (self.sigma * sqrt(2.0)))) + + @property + def variance(self): + 'Square of the standard deviation' + return self.sigma ** 2.0 + + def __add__(x1, x2): + if isinstance(x2, NormalDist): + return NormalDist(x1.mu + x2.mu, hypot(x1.sigma, x2.sigma)) + return NormalDist(x1.mu + x2, x1.sigma) + + def __sub__(x1, x2): + if isinstance(x2, NormalDist): + return NormalDist(x1.mu - x2.mu, hypot(x1.sigma, x2.sigma)) + return NormalDist(x1.mu - x2, x1.sigma) + + def __mul__(x1, x2): + return NormalDist(x1.mu * x2, x1.sigma * fabs(x2)) + + def __truediv__(x1, x2): + return NormalDist(x1.mu / x2, x1.sigma / fabs(x2)) + + def __pos__(x1): + return x1 + + def __neg__(x1): + return NormalDist(-x1.mu, x1.sigma) + + __radd__ = __add__ + + def __rsub__(x1, x2): + return -(x1 - x2) + + __rmul__ = __mul__ + + def __eq__(x1, x2): + if not isinstance(x2, NormalDist): + return NotImplemented + return (x1.mu, x2.sigma) == (x2.mu, x2.sigma) + + def __repr__(self): + return f'{type(self).__name__}(mu={self.mu!r}, sigma={self.sigma!r})' + + +if __name__ == '__main__': + + # Show math operations computed analytically in comparsion + # to a monte carlo simulation of the same operations + + from math import isclose + from operator import add, sub, mul, truediv + from itertools import repeat + + g1 = NormalDist(10, 20) + g2 = NormalDist(-5, 25) + + # Test scaling by a constant + assert (g1 * 5 / 5).mu == g1.mu + assert (g1 * 5 / 5).sigma == g1.sigma + + n = 100_000 + G1 = g1.samples(n) + G2 = g2.samples(n) + + for func in (add, sub): + print(f'\nTest {func.__name__} with another NormalDist:') + print(func(g1, g2)) + print(NormalDist.from_samples(map(func, G1, G2))) + + const = 11 + for func in (add, sub, mul, truediv): + print(f'\nTest {func.__name__} with a constant:') + print(func(g1, const)) + print(NormalDist.from_samples(map(func, G1, repeat(const)))) + + const = 19 + for func in (add, sub, mul): + print(f'\nTest constant with {func.__name__}:') + print(func(const, g1)) + print(NormalDist.from_samples(map(func, repeat(const), G1))) + + def assert_close(G1, G2): + assert isclose(G1.mu, G1.mu, rel_tol=0.01), (G1, G2) + assert isclose(G1.sigma, G2.sigma, rel_tol=0.01), (G1, G2) + + X = NormalDist(-105, 73) + Y = NormalDist(31, 47) + s = 32.75 + n = 100_000 + + S = NormalDist.from_samples([x + s for x in X.samples(n)]) + assert_close(X + s, S) + + S = NormalDist.from_samples([x - s for x in X.samples(n)]) + assert_close(X - s, S) + + S = NormalDist.from_samples([x * s for x in X.samples(n)]) + assert_close(X * s, S) + + S = NormalDist.from_samples([x / s for x in X.samples(n)]) + assert_close(X / s, S) + + S = NormalDist.from_samples([x + y for x, y in zip(X.samples(n), + Y.samples(n))]) + assert_close(X + Y, S) + + S = NormalDist.from_samples([x - y for x, y in zip(X.samples(n), + Y.samples(n))]) + assert_close(X - Y, S) diff --git a/Lib/test/test_statistics.py b/Lib/test/test_statistics.py index e35144677ad..a65fbe8dd25 100644 --- a/Lib/test/test_statistics.py +++ b/Lib/test/test_statistics.py @@ -5,9 +5,11 @@ approx_equal function. import collections import collections.abc +import copy import decimal import doctest import math +import pickle import random import sys import unittest @@ -2025,6 +2027,181 @@ class TestStdev(VarianceStdevMixin, NumericTestCase): expected = math.sqrt(statistics.variance(data)) self.assertEqual(self.func(data), expected) +class TestNormalDist(unittest.TestCase): + + def test_slots(self): + nd = statistics.NormalDist(300, 23) + with self.assertRaises(TypeError): + vars(nd) + self.assertEqual(nd.__slots__, ('mu', 'sigma')) + + def test_instantiation_and_attributes(self): + nd = statistics.NormalDist(500, 17) + self.assertEqual(nd.mu, 500) + self.assertEqual(nd.sigma, 17) + self.assertEqual(nd.variance, 17**2) + + # default arguments + nd = statistics.NormalDist() + self.assertEqual(nd.mu, 0) + self.assertEqual(nd.sigma, 1) + self.assertEqual(nd.variance, 1**2) + + # error case: negative sigma + with self.assertRaises(statistics.StatisticsError): + statistics.NormalDist(500, -10) + + def test_alternative_constructor(self): + NormalDist = statistics.NormalDist + data = [96, 107, 90, 92, 110] + # list input + self.assertEqual(NormalDist.from_samples(data), NormalDist(99, 9)) + # tuple input + self.assertEqual(NormalDist.from_samples(tuple(data)), NormalDist(99, 9)) + # iterator input + self.assertEqual(NormalDist.from_samples(iter(data)), NormalDist(99, 9)) + # error cases + with self.assertRaises(statistics.StatisticsError): + NormalDist.from_samples([]) # empty input + with self.assertRaises(statistics.StatisticsError): + NormalDist.from_samples([10]) # only one input + + def test_sample_generation(self): + NormalDist = statistics.NormalDist + mu, sigma = 10_000, 3.0 + X = NormalDist(mu, sigma) + n = 1_000 + data = X.samples(n) + self.assertEqual(len(data), n) + self.assertEqual(set(map(type, data)), {float}) + # mean(data) expected to fall within 8 standard deviations + xbar = statistics.mean(data) + self.assertTrue(mu - sigma*8 <= xbar <= mu + sigma*8) + + # verify that seeding makes reproducible sequences + n = 100 + data1 = X.samples(n, seed='happiness and joy') + data2 = X.samples(n, seed='trouble and despair') + data3 = X.samples(n, seed='happiness and joy') + data4 = X.samples(n, seed='trouble and despair') + self.assertEqual(data1, data3) + self.assertEqual(data2, data4) + self.assertNotEqual(data1, data2) + + # verify that subclass type is honored + class NewNormalDist(NormalDist): + pass + nnd = NewNormalDist(200, 5) + self.assertEqual(type(nnd), NewNormalDist) + + def test_pdf(self): + NormalDist = statistics.NormalDist + X = NormalDist(100, 15) + # Verify peak around center + self.assertLess(X.pdf(99), X.pdf(100)) + self.assertLess(X.pdf(101), X.pdf(100)) + # Test symmetry + self.assertAlmostEqual(X.pdf(99), X.pdf(101)) + self.assertAlmostEqual(X.pdf(98), X.pdf(102)) + self.assertAlmostEqual(X.pdf(97), X.pdf(103)) + # Test vs CDF + dx = 2.0 ** -10 + for x in range(90, 111): + est_pdf = (X.cdf(x + dx) - X.cdf(x)) / dx + self.assertAlmostEqual(X.pdf(x), est_pdf, places=4) + # Error case: variance is zero + Y = NormalDist(100, 0) + with self.assertRaises(statistics.StatisticsError): + Y.pdf(90) + + def test_cdf(self): + NormalDist = statistics.NormalDist + X = NormalDist(100, 15) + cdfs = [X.cdf(x) for x in range(1, 200)] + self.assertEqual(set(map(type, cdfs)), {float}) + # Verify montonic + self.assertEqual(cdfs, sorted(cdfs)) + # Verify center + self.assertAlmostEqual(X.cdf(100), 0.50) + # Error case: variance is zero + Y = NormalDist(100, 0) + with self.assertRaises(statistics.StatisticsError): + Y.cdf(90) + + def test_same_type_addition_and_subtraction(self): + NormalDist = statistics.NormalDist + X = NormalDist(100, 12) + Y = NormalDist(40, 5) + self.assertEqual(X + Y, NormalDist(140, 13)) # __add__ + self.assertEqual(X - Y, NormalDist(60, 13)) # __sub__ + + def test_translation_and_scaling(self): + NormalDist = statistics.NormalDist + X = NormalDist(100, 15) + y = 10 + self.assertEqual(+X, NormalDist(100, 15)) # __pos__ + self.assertEqual(-X, NormalDist(-100, 15)) # __neg__ + self.assertEqual(X + y, NormalDist(110, 15)) # __add__ + self.assertEqual(y + X, NormalDist(110, 15)) # __radd__ + self.assertEqual(X - y, NormalDist(90, 15)) # __sub__ + self.assertEqual(y - X, NormalDist(-90, 15)) # __rsub__ + self.assertEqual(X * y, NormalDist(1000, 150)) # __mul__ + self.assertEqual(y * X, NormalDist(1000, 150)) # __rmul__ + self.assertEqual(X / y, NormalDist(10, 1.5)) # __truediv__ + with self.assertRaises(TypeError): + y / X + + def test_equality(self): + NormalDist = statistics.NormalDist + nd1 = NormalDist() + nd2 = NormalDist(2, 4) + nd3 = NormalDist() + self.assertNotEqual(nd1, nd2) + self.assertEqual(nd1, nd3) + + # Test NotImplemented when types are different + class A: + def __eq__(self, other): + return 10 + a = A() + self.assertEqual(nd1.__eq__(a), NotImplemented) + self.assertEqual(nd1 == a, 10) + self.assertEqual(a == nd1, 10) + + # All subclasses to compare equal giving the same behavior + # as list, tuple, int, float, complex, str, dict, set, etc. + class SizedNormalDist(NormalDist): + def __init__(self, mu, sigma, n): + super().__init__(mu, sigma) + self.n = n + s = SizedNormalDist(100, 15, 57) + nd4 = NormalDist(100, 15) + self.assertEqual(s, nd4) + + # Don't allow duck type equality because we wouldn't + # want a lognormal distribution to compare equal + # to a normal distribution with the same parameters + class LognormalDist: + def __init__(self, mu, sigma): + self.mu = mu + self.sigma = sigma + lnd = LognormalDist(100, 15) + nd = NormalDist(100, 15) + self.assertNotEqual(nd, lnd) + + def test_pickle_and_copy(self): + nd = statistics.NormalDist(37.5, 5.625) + nd1 = copy.copy(nd) + self.assertEqual(nd, nd1) + nd2 = copy.deepcopy(nd) + self.assertEqual(nd, nd2) + nd3 = pickle.loads(pickle.dumps(nd)) + self.assertEqual(nd, nd3) + + def test_repr(self): + nd = statistics.NormalDist(37.5, 5.625) + self.assertEqual(repr(nd), 'NormalDist(mu=37.5, sigma=5.625)') + # === Run tests === diff --git a/Misc/NEWS.d/next/Library/2019-02-21-15-47-00.bpo-36018.qt7QUe.rst b/Misc/NEWS.d/next/Library/2019-02-21-15-47-00.bpo-36018.qt7QUe.rst new file mode 100644 index 00000000000..bba47f4ea91 --- /dev/null +++ b/Misc/NEWS.d/next/Library/2019-02-21-15-47-00.bpo-36018.qt7QUe.rst @@ -0,0 +1,3 @@ +Add statistics.NormalDist, a tool for creating and manipulating normal +distributions of random variable. Features a composite class that treats +the mean and standard deviation of measurement data as single entity.