Fuzz test for histogram consistency.

test_fuzzed_histogram.py

import numpy as np


class BaseHistogramAccuracy(object):
    # The number of fuzz tests to perform.
    n_fuzz = 1000

    # The PRNG seed to use.
    seed = 8675309

    def linspace(self, low, high, steps):
        # Override with your best linspace function here.
        raise NotImplementedError

    def test_histogram_accuracy(self):
        prng = np.random.RandomState(self.seed)
        for i in range(self.n_fuzz):
            low = float(prng.randint(-100, 101))
            width = prng.randint(100, 2000)
            high = low + width
            nbins = width * prng.randint(1, 20)
            yield self.check_accuracy, low, high, nbins

    def check_accuracy(self, low, high, nbins):
        steps = nbins + 1
        edges = self.linspace(low, high, steps)
        inner = edges[1:-1]
        prev_inner = np.nextafter(inner, -np.inf)
        next_inner = np.nextafter(inner, np.inf)
        # The number just prior to each edge should be in the bin below said
        # edge.
        np.testing.assert_array_equal(
            np.histogram(prev_inner, range=(low, high), bins=nbins)[0],
            np.hstack([1] * len(inner) + [0]))
        # The edge itself should fall in the bin above the edge.
        np.testing.assert_array_equal(
            np.histogram(inner, range=(low, high), bins=nbins)[0],
            np.hstack([0] + [1] * len(inner)))
        # And so should the number immediately after the edge.
        np.testing.assert_array_equal(
            np.histogram(next_inner, range=(low, high), bins=nbins)[0],
            np.hstack([0] + [1] * len(inner)))

class TestLinspaceHistogramAccuracy(BaseHistogramAccuracy):
    def linspace(self, low, high, steps):
        x = np.linspace(low, high, steps)
        return x