onstats 0.9.4

Online stats in python, based in generators with send

Generators Based Online Statistics

This package implement online statistics written using python generators, with the only depencency of numpy.

It can calculate basic stats in an online manner with good stability.

The Repo focus in simplicity and compousability, reliying 100% on python Generators.

By desingn window operations increase in window size till the desired size instead of returning Nan in the initialization period as pd.rolling does.

install

poetry add onstats

pip install onstats

How to use:

Import the window / rolling statistic you want to compute and send the values to it:

>>>  from onstats.stats import ma # moving average
>>>  from onstats.util import send

>>>  gma = ma(2)  # with window 2
>>>  gma.send(3)
3
>>>  gma.send(5)
4
>>>  gma.send(5)
5

If w = 0 the window is infinitelly large , we will compute the normal average.

You can also feed all the iterator directly:

>>>  gma = ma(2)  # with window 2
>>>  send(gma, [3,5,5])
5

Or as an iterator

>>>  gma = ma(2)  # with window 2
>>>  for d in isend(gma, [3,5,5]):
>>>     print(d)
3
4
5

You can also pass 2d np.arrays:

>>>  gma = ma(2)  # with window 2
>>>  for d in isend(gma, np.array([[0,0],[1,2],[1,4]])):
>>>     print(d)
[0,0]
[0.5,1]
[1,3]

Supported Stats:

rolling	window	infinite	Ddoff	Description
ma	✅	✅		Moving Average
ema		✅		Exponential Moving Average
var	✅	✅	✅	Variance
ath		✅		All Time High
wsum	✅	✅		Windowed Sum
cov_xy	✅		✅	Covariance
corr_xy	✅		✅	Correlation
auto_corr	✅		✅	Auto Correlation