如何使用列表设置scipy.stats分发的参数

Question

I am using the package fitter to fit continuous distributions to numpy arrays of data in Python. fitter.fitted_param returns a dictionary of tupples of varying length with the best-fit parameters for each distribution. These are the same parameters used to set these distributions in scipy.stats . I would like to use these tuples to set the parameters directly with scipy.stats , but I do not know how to do so. Any ideas?

Example:

>from fitter import Fitter
>import numpy as np
>data = np.random.random((1000,1))
>f = Fitter(data,distributions = ['norm','gamma'])
>f.fit()
>param = f.fitted_param['gamma']
>param
out:(20759.430545279687, -41.012521759919224, 0.0019996776498165851)

If I now want to create a new gamma distribution to generate gamma-distributed random values with the same a , loc and scale specified in param above, I need to write:

>from scipy import stats
>rv = stats.gamma(a=param[0],loc=param[1],scale=param[2])
>rv.rvs(100)

How can I set a , loc ,and scale directly without typing them out explicitly? I want to be able to quickly fit any distribution with an arbitrary number of parameters and generate random values that follow that distribution.

Thanks for your help!

Answer 1

You can do <distribution>(*param) , provided that the list of parameters param is in the right order.

If it is not, you have to have your params in a dictionary and call it as keyword arguments with <distribution>(**param) , for instance:

param = {"a": 20759, "scale": 0.001, "loc": -41}
stats.gamma(**param)

Edit: more information here: *args and **kwargs?