This is a fork from the original project by @tsakim This version adds these Features:
- Adds
mean,var,std,skew,amaxandargmaxmethods - Solves FFT instability issues, as proposed here by @zeeshansayyed
The module contains a Python implementation of functions related to the Poisson Binomial probability distribution [1], which describes the probability distribution of the sum of independent Bernoulli random variables with non-uniform success probabilities. For further information, see reference [1].
The implemented methods are:
pmf: probability mass functioncdf: cumulative distribution functionpval: p-value for right tailed testsmean: mean of the distributionvar: variance of the distributionstd: standard deviation of the distributionskew: skewness of the distributionamax: max value of the probability mass functionargmax: index of the max value of the probability mass function
Mika Straka, Maxi Marufo, Zeeshan Sayyed
Consider n independent and non-identically distributed random variables and
be p a list/NumPy array of the corresponding Bernoulli success probabilities.
In order to create the Poisson Binomial distributions, use
from poibin import PoiBin
pb = PoiBin(p)- Mean
pb.mean()- Variance
pb.var()- Standard Deviation
pb.std()- Skewness
pb.skew()- Max value
pb.amax()- Index of the max value
pb.argmax()Be x a list/NumPy array of different numbers of success. Use the following
methods to obtain the corresponding quantities:
- Probability mass function
pb.pmf(x)- Cumulative distribution function
pb.cdf(x)- P-values for right tailed tests
pb.pval(x)All three methods accept single integers as well as lists/NumPy arrays of
integers. Note that x[i] must be smaller than len(p).
The methods have been implemented using the pytest module. To run the tests, execute
$ pytest test_poibin.py
in the command line. For verbose mode, use
$ pytest -v test_poibin.py
Copyright (c) 2016-2020 Mika J. Straka, Maxi Marufo, Zeeshan Sayyed