def test_mean_doesnt_mutate_argument(self):
xs = [-2.9, -64.34, 98.76, 56.14, 20.62, 120.62, 22.33, 277.13, 45.1,
-80.51, 114.01, 209.34, -79.91, 200.83, 144.65, -53.38, -92.95,
208.1, 84.47, 138.65]
copy = xs[:]
basic_stats.mean(xs)
assert_equals(xs, copy)
def test_mean_doesnt_mutate_argument(self):
xs = [
-2.9, -64.34, 98.76, 56.14, 20.62, 120.62, 22.33, 277.13, 45.1,
-80.51, 114.01, 209.34, -79.91, 200.83, 144.65, -53.38, -92.95,
208.1, 84.47, 138.65
]
copy = xs[:]
basic_stats.mean(xs)
assert_equals(xs, copy)
def test_mean_return_value(self):
xs = [
-2.9, -64.34, 98.76, 56.14, 20.62, 120.62, 22.33, 277.13, 45.1,
-80.51, 114.01, 209.34, -79.91, 200.83, 144.65, -53.38, -92.95,
208.1, 84.47, 138.65
]
assert_equals(basic_stats.mean(xs), 68.338)
def scottknott(data, max_rank_size=3, epsilon=0.01):
"""
Recursively split data, maximizing delta of the expected value of the
mean before and after the splits. Reject splits with under max_rank_size
items.
"""
flat_data = [x for log in data for x in log.contents()]
data_mean = basic_stats.mean(flat_data)
def recurse_and_rank(parts, rank=0):
"Split, then recurse_and_rank on each part."
cut = min_mu(parts, data_mean, len(flat_data), max_rank_size, epsilon)
if cut:
# if cut, rank "right" higher than "left"
rank = recurse_and_rank(parts[:cut], rank) + 1
rank = recurse_and_rank(parts[cut:], rank)
else:
# if no cut, then all get same rank
for part in parts:
part.rank = rank
return rank
recurse_and_rank(sorted(data, key=lambda x: x.median()))
return data
def mean(self):
if self._cached_mean is not None:
return self._cached_mean
self._cached_mean = basic_stats.mean(self._cache)
return self._cached_mean
def test_mean_return_value(self):
xs = [-2.9, -64.34, 98.76, 56.14, 20.62, 120.62, 22.33, 277.13, 45.1,
-80.51, 114.01, 209.34, -79.91, 200.83, 144.65, -53.38, -92.95,
208.1, 84.47, 138.65]
assert_equals(basic_stats.mean(xs), 68.338)
def mean(self):
if self._cached_mean is not None:
return self._cached_mean
self._cached_mean = basic_stats.mean(self._cache)
return self._cached_mean
