How to random sample a 2-parameter weibull distrib

问题:

I was wondering how to generate a random weibull distribution with 2-parameter (lambda, k) in python. I know that numpy has a numpy.random.weibull, but it only accepts the a parameter as the shape of the distribution.

回答1:

Severin Pappadeux's answer is probably the simplest way to include the scale parameter. An alternative is to use scipy.stats.weibull_min. weibull_min has three parameters: shape, location and scale. You only want the shape and scale, so you would set the location to 0.

from scipy.stats import weibull_min

n = 100     # number of samples
k = 2.4     # shape
lam = 5.5   # scale

x = weibull_min.rvs(k, loc=0, scale=lam, size=n)

回答2:

Well, if you sample a number from weibull distribution with scale parameter missing (which assumes scale is equal to 1), then to get it scale multiply by lambda.

x = numpy.random.weibull(a)
return lambda*x