How to create a discrete distribution in OpenTURNS?

Question

I have a sample of real values which contains independent realizations of a discrete random variable and I want to create the distribution which fits this data.

sample = [2.0, 2.0, 1.0, 1.0, 2.0, 3.0, 1.0, 2.0, 2.0, 1.0]

The UserDefined distribution seems to be designed for this purpose, but requires to compute the weights of each point, depending on its frequency in the sample:

import openturns as ot
distribution = ot.UserDefined(points, weights)

But we have to compute the points and weights first. To do this, I computed the points and weights using the Numpy unique function. However, this sounds like a limitation of the UserDefined class. How may I do this more simply?

Answer 1

The UserDefinedFactory class creates a UserDefined distribution by estimating the points and weights from the sample. The build method takes the sample as input and returns the ot.UserDefined object that fits the data.

import openturns as ot
sample = ot.Sample([[2.0], [2.0], [1.0], [1.0], [2.0], [3.0], [1.0], [2.0], [2.0], [1.0]])
factory = ot.UserDefinedFactory()
distribution = factory.build(sample)