CODEWITHSUNDEEP
pythonnumpypandasscipy
Baktaawar
Baktaawar

Reputation: 7500

Generating Multivariate Distributions(data points) in Python

I have a couple of questions:

1). If I need to generate a 1000x100 data points(dataset), with each co-ordinate having mean 0 and 1(marginal distribution), what's the best way to do that in numpy or pandas if possible?

2). Generate a multivariate data points 100x1000.

I can do these by creating 100 list of 1000 dimensions each with a univariate distribution(mean 0 and 1) and can then concat it together in numpy. But I think that would be a very long approach. I am looking for a more pythonic way. Also I would appreciate if a general approach could be given which could help me generate any distribution for a NxD dimensional dataset.

Upvotes: 1

Views: 481

Answers (1)

cphlewis
cphlewis

Reputation: 16269

Numpy's random number generators will fill any shape you want:

import numpy.random as nr
nr.seed()
nr.normal(0,1,(2,3,4))

Upvotes: 1

Related Questions