def test_addition():
hist_a = traces.Histogram([1, 1, 1, 2, 3, 5])
hist_b = traces.Histogram([0, 0, 1, 2, 2])
together = hist_a.add(hist_b)
assert list(together.items()) == [(0, 2), (1, 4), (2, 3), (3, 1), (5, 1)]
def test_normalize():
data = [15, 15, 20, 20, 20, 35, 35, 40, 40, 50, 50]
histogram = traces.Histogram(data)
normalized = histogram.normalized()
assert sum(normalized.itervalues()) == 1.0
def test_quantile_interpolation():
data = [1, 1, 1, 2, 3, 5, 6, 7]
histogram = traces.Histogram(data)
normalized = histogram.normalized()
q_list = [0.001, 0.01, 0.05, 0.25, 0.5, 0.75, 0.95, 0.99, 0.999]
# just do the inverse of the emperical cdf
result = histogram.quantiles(q_list, alpha=0, smallest_count=1)
answer = [1.0, 1.0, 1.0, 1.0, 2.5, 5.5, 7.0, 7.0, 7.0]
for i, j in zip(result, answer):
nose.tools.assert_almost_equal(i, j)
# same thing with normalized
result = normalized.quantiles(
q_list, alpha=0, smallest_count=1.0 / len(data))
for i, j in zip(result, answer):
nose.tools.assert_almost_equal(i, j)
# now do the linear interpolation method
result = histogram.quantiles(q_list, alpha=0.5, smallest_count=1)
answer = stats.mstats.mquantiles(
data, prob=q_list, alphap=0.5, betap=0.5,
)
for i, j in zip(result, answer):
nose.tools.assert_almost_equal(i, j)
# same thing with normalized
result = normalized.quantiles(
q_list, alpha=0.5, smallest_count=1.0 / len(data),
)
for i, j in zip(result, answer):
nose.tools.assert_almost_equal(i, j)
def test_histogram_stats_with_nones():
histogram = traces.Histogram()
nose.tools.eq_(histogram.mean(), None)
nose.tools.eq_(histogram.variance(), None)
nose.tools.eq_(histogram.standard_deviation(), None)
nose.tools.eq_(histogram.min(), None)
nose.tools.eq_(histogram.max(), None)
nose.tools.eq_(histogram.median(), None)
histogram = traces.Histogram.from_dict({None: 1}, key=hash)
nose.tools.eq_(histogram.mean(), None)
nose.tools.eq_(histogram.variance(), None)
nose.tools.eq_(histogram.standard_deviation(), None)
nose.tools.eq_(histogram.min(), None)
nose.tools.eq_(histogram.max(), None)
nose.tools.eq_(histogram.median(), None)
ts = traces.TimeSeries()
ts[0] = None
ts[1] = 5
ts[2] = 6
ts[3] = None
ts[9] = 7
ts[10] = None
histogram = ts.distribution(start=0, end=10)
nose.tools.eq_(histogram.mean(), 6)
def test_minmax_with_zeros():
histogram = traces.Histogram()
histogram[0] += 0
histogram[1] += 1
histogram[2] += 1
histogram[3] += 0
nose.tools.eq_(histogram.min(), 1)
nose.tools.eq_(histogram.max(), 2)
def test_quantiles():
data = [15, 15, 20, 20, 20, 35, 35, 40, 40, 50, 50]
histogram = traces.Histogram(data)
alpha = 0.5
q_list = [0.05, 0.25, 0.5, 0.75, 0.95]
q_values = histogram.quantiles(q_list, alpha=alpha, smallest_count=1)
reference = \
stats.mstats.mquantiles(data, prob=q_list, alphap=0.5, betap=0.5)
for i, j in zip(q_values, reference):
assert i == j
def _test_statistics(normalized):
data_list = [
[1, 2, 3, 5, 6, 7],
[1, 2, 3, 5, 6],
[1, 1],
[1, 1, 1, 1, 1, 1, 1, 2],
[i + 0.25 for i in [1, 1, 1, 1, 1, 1, 1, 2]],
]
for data in data_list:
histogram = traces.Histogram(data)
if normalized:
histogram = histogram.normalized()
n = 1
else:
n = len(data)
nose.tools.assert_almost_equal(histogram.total(), n)
nose.tools.assert_almost_equal(histogram.mean(), numpy.mean(data))
nose.tools.assert_almost_equal(histogram.variance(), numpy.var(data))
nose.tools.assert_almost_equal(
histogram.standard_deviation(),
numpy.std(data),
)
nose.tools.assert_almost_equal(histogram.max(), numpy.max(data))
nose.tools.assert_almost_equal(histogram.min(), numpy.min(data))
nose.tools.assert_almost_equal(
histogram.quantile(0.5),
numpy.median(data),
)
q_list = [0.001, 0.01, 0.05, 0.25, 0.5, 0.75, 0.95, 0.99, 0.999]
# linear interpolation
result = histogram.quantiles(q_list)
reference = stats.mstats.mquantiles(
data,
prob=q_list,
alphap=0.5,
betap=0.5,
)
for i, j in zip(result, reference):
nose.tools.assert_almost_equal(i, j)
# make sure ot throw an error for bad quantile values
try:
histogram.quantile(-1)
except ValueError:
pass
