views:

632

answers:

2

Are there any built-in libraries in Python or Numpy to generate random numbers based on various common distributions, such as:

  • Normal
  • Poisson
  • Exponential
  • Bernoulli

And various others?

Are there any such libraries with multi-variate distributions?

+3  A: 

The random module has tons of functions for generating random numbers in lots of way. Not sure it has multi-variate.

Numpy.random would be the next place to look.

Ned Batchelder
I took the liberty of adding in a link to the documentation for `random`
Nathan Fellman
+10  A: 
#!/usr/bin/env python
from scipy.stats import bernoulli,poisson,norm,expon

bernoulli, poisson, norm, expon and many others are documented here

print(norm.rvs(size=30))
print(bernoulli.rvs(.3,size=30))
print(poisson.rvs(1,2,size=30))
print(expon.rvs(5,size=30))

All the distributions defined in scipy.stats have a common interface to the pdf, cdf, rvs (random variates). More info here.

unutbu
Import soul. Damn you Python !
e-satis