"""
This module accumulates statistical values to analyze its mean, standard deviation and moments.
"""
import numpy as np
[docs]class Accumulator:
"""Accumulate values."""
def __init__(self, num_moments=3, values=None):
self._moments = np.zeros(num_moments)
if values is not None:
self.add(values)
[docs] def num_moments(self):
"""Return the number of moments."""
return len(self._moments)
[docs] def add(self, value):
"""Add a value or values to the moments."""
if isinstance(value, list):
for val in value:
self.add(val)
else:
mom = 1
for index, _ in enumerate(self._moments):
self._moments[index] += mom
mom *= value
[docs] def num_values(self):
"""Return the number of values."""
return self._moments[0]
[docs] def mean(self):
"""
Return the mean.
>>> from pyfeasst import accumulator
>>> acc = accumulator.Accumulator()
>>> acc.add(1.)
>>> acc.add(2.)
>>> acc.add(4.)
>>> acc.add([3, 5, 7])
>>> acc.mean()
3.6666666666666665
"""
mean = float('nan')
if self.num_values() > 0 and self.num_moments() > 1:
mean = self.sum_moment(1)/self.num_values()
return mean
[docs] def stdev(self):
"""
Return the standard deviation.
>>> from pyfeasst import accumulator
>>> acc = accumulator.Accumulator(values=[1, 2, 3, 4, 5, 7])
>>> acc.stdev()
2.160246899469287
"""
std = float('nan')
if self.num_values() > 2 and self.num_moments() > 2:
num = self.num_values()
std = np.sqrt((self.sum_moment(2)/num - (self.sum_moment(1)/num)**2)*num/(num-1))
return std
[docs] def sum_moment(self, power):
"""
Return the sum of the moment, x**power.
>>> from pyfeasst import accumulator
>>> acc = accumulator.Accumulator(values=[1, 2, 3, 4, 5, 7])
>>> acc.sum_moment(0)
6.0
>>> acc.sum_moment(1)
22.0
>>> acc.sum_moment(2)
104.0
"""
mom = float('nan')
if self.num_values() > 0 and self.num_moments() > power:
mom = self._moments[power]
return mom
if __name__ == "__main__":
import doctest
doctest.testmod()