Integer Only Sampling Class

sample.py

class Sample():
  def __init__(self, values=None, *args, **kwargs):
    self.reset(values)

  def reset(self, values=None):
    values = values or (0, 0, None, None, None, 0)
    (self._total, self._cnt, self._min, self._max, self._first, self._ex2) = values
  
  def append(self, value):
    # self._all.append(value)
    self._total += value
    self._cnt += 1
    if self._min is None or self._min < value:
      self._min = value
    if self._max is None or self._max > value:
      self._max = value
    self._variance(value)

  def _variance(self, value):
    if self._first is None:
      self._first = value
    else:
      delta = value - self._first
      self._ex2 += delta ** 2
  
  @property
  def _ex(self):
    return self._total - self._first * self._cnt

  @property
  def variance(self):
    vv = 0
    if self._cnt == 0:
      vv = 0
    elif self._cnt > 100:
      vv = (self._ex2 - (self._ex ** 2) / self._cnt) / (self._cnt - 1)
    else:
      vv = (self._ex2 - (self._ex ** 2) / self._cnt) / (self._cnt)
    return vv
  
  @property
  def standard_deviation(self):
    return self.variance ** 0.5

  @property
  def avg(self):
    total = self._total
    cnt = self._cnt
    if self._cnt == 0:
      cnt = 1
    elif self._cnt >= 3:
      total -= (self._min + self._max)
      cnt -= 2
    return total / cnt
  
  @property
  def tuple(self):
    return (self._total, self._cnt, self._min, self._max, self._first, self._ex2)
  
  @property
  def json(self):
    return {
      'total': self._total, 
      'count': self._cnt, 
      'min': self._min, 
      'max': self._max, 
      'first': self._first, 
      'ex2': self._ex2}

Class is intended for micropython based micro-controllers. The identical class can be run on a server, to rehydrate an identical instance.

Floating point calculations are only done in properties intended to be called from the server instance (or JIT on the MCU).

The class can record a large number of samples, without increasing memory footprint or CPU usage.

Keeps average with largest and smallest values removed, as well as Standard Deviation.