def unweighted_nine(distribution_or_values):
"""
Returns the unweighted "Nines" box-whisker boundaries for the given
distribution, including the median, where the whiskers are defined by the
9% and 91% percentiles.
:param distribution_or_values:
The distribution for which to calculate the boundaries
:return:
A tuple containing 5 values:
* minimum (whisker boundary)
* lower quartile (box boundary)
* median
* upper quartile (box boundary)
* maximum (whisker boundary)
:rtype: tuple
"""
values = support.to_unweighted_population(distribution_or_values)
return (
np.percentile(values, 9),
np.percentile(values, 25),
np.percentile(values, 50),
np.percentile(values, 75),
np.percentile(values, 91)
)
def unweighted_tukey(distribution_or_values):
"""
Returns the unweighted Tukey box-whisker boundaries for the given
distribution, including the median. As the returned boundaries are
unweighted, they ignore uncertainties in the measurements.
:param distribution_or_values:
The distribution for which to calculate the boundaries
:return:
A tuple containing 5 values:
* minimum (whisker boundary)
* lower quartile (box boundary)
* median
* upper quartile (box boundary)
* maximum (whisker boundary)
:rtype: tuple
"""
values = support.to_unweighted_population(distribution_or_values)
median = np.percentile(values, 50)
lower_quartile = np.percentile(values, 25)
upper_quartile = np.percentile(values, 75)
inter_quartile_range = upper_quartile - lower_quartile
tukey_range = 1.5 * inter_quartile_range
values.sort()
bottom = None
top = None
for index in range(len(values)):
v = values[index]
if bottom is None and v >= (median - tukey_range):
bottom = v
v = values[len(values) - index - 1]
if top is None and v <= (median + tukey_range):
top = v
if top is not None and bottom is not None:
break
return bottom, lower_quartile, median, upper_quartile, top
def unweighted_tukey(distribution_or_values):
"""
Returns the unweighted Tukey box-whisker boundaries for the given
distribution, including the median. As the returned boundaries are
unweighted, they ignore uncertainties in the measurements.
:param distribution_or_values:
The distribution for which to calculate the boundaries
:return:
A tuple containing 5 values:
* minimum (whisker boundary)
* lower quartile (box boundary)
* median
* upper quartile (box boundary)
* maximum (whisker boundary)
:rtype: tuple
"""
values = support.to_unweighted_population(distribution_or_values)
median = np.percentile(values, 50)
lower_quartile = np.percentile(values, 25)
upper_quartile = np.percentile(values, 75)
inter_quartile_range = upper_quartile - lower_quartile
tukey_range = 1.5 * inter_quartile_range
values.sort()
bottom = None
top = None
for index in range(len(values)):
v = values[index]
if bottom is None and v >= (median - tukey_range):
bottom = v
v = values[len(values) - index - 1]
if top is None and v <= (median + tukey_range):
top = v
if top is not None and bottom is not None:
break
return bottom, lower_quartile, median, upper_quartile, top
def unweighted_nine(distribution_or_values):
"""
Returns the unweighted "Nines" box-whisker boundaries for the given
distribution, including the median, where the whiskers are defined by the
9% and 91% percentiles.
:param distribution_or_values:
The distribution for which to calculate the boundaries
:return:
A tuple containing 5 values:
* minimum (whisker boundary)
* lower quartile (box boundary)
* median
* upper quartile (box boundary)
* maximum (whisker boundary)
:rtype: tuple
"""
values = support.to_unweighted_population(distribution_or_values)
return (np.percentile(values, 9), np.percentile(values, 25),
np.percentile(values,
50), np.percentile(values,
75), np.percentile(values, 91))
