stats.rst

Extended doctests for stats

stats are computed quickly, even over 1 million samples:

>>> from time import perf_counter as clock
>>> start = clock()
>>> s = stats(range(1_000_000))
>>> s.pprint()
stats(
    count=1000000,
    first=0,
    last=999999,
    min=0,
    max=999999,
    sum=499999500000,
    mean=499999.5,
    ssdm=8.333333333324998e+16,
    # variance: 83333416666.66666
    # stdev: 288675.27893234405
)
>>> end = clock()
>>> assert end - start < 0.5

Addition works as expected:

>>> s = stats([0])
>>> s
stats(count=1, first=0, last=0, min=0, max=0, sum=0, mean=0, ssdm=0)

>>> s + [1]
stats(count=2, first=0, last=1, min=0, max=1, sum=1, mean=0.5, ssdm=0.5)

>>> s += [1]
>>> s
stats(count=2, first=0, last=1, min=0, max=1, sum=1, mean=0.5, ssdm=0.5)

Works with sum():

>>> sum([[2], [3]], s)
stats(count=4, first=0, last=3, min=0, max=3, sum=6, mean=1.5, ssdm=5.0)

In-place addition reassigns, and doesn't mutate:

>>> s0 = s1 = stats([0])
>>> assert s1 is s0
>>> assert s1 == s0
>>> s1 += [0]
>>> assert s1 is not s0  # s1 has been reassigned.
>>> assert s1 != s0  # Its value has changed.

Empty samples don't change any values:

>>> s0 = s1 = stats([0])
>>> assert s1 is s0
>>> assert s1 == s0
>>> s1 += ()
>>> assert s1 is not s0  # s1 has been reassigned.
>>> assert s1 == s0  # Its value has *not* changed.
>>> assert repr(s1) == repr(s0)  # mean and ssdm are still ints.

Miscellaneous corner cases:

>>> stats([0]).pprint()
stats(
    count=1,
    first=0,
    last=0,
    min=0,
    max=0,
    sum=0,
    mean=0,
    ssdm=0,
    # variance: 0.0
    # stdev: 0.0
)

>>> stats([1]).pprint()
stats(
    count=1,
    first=1,
    last=1,
    min=1,
    max=1,
    sum=1,
    mean=1,
    ssdm=0,
    # variance: 0.0
    # stdev: 0.0
)

>>> stats([])
Traceback (most recent call last):
  ...
StopIteration