1

Is there a way to make random draws from a lognormal distribution with a skewed shape parameter?

There are a few questions currently leaving something to be desired on the same topic here & here.

I'm currently using scipy.stats.lognorm. This accepts shape, scale, and location. So, location is the mean (expeceted value), scale is the standard deviation (or spread of the dist), and shape is the skew (?).

How is the skew, typically γ, incorporated into this distribution? Now I can currently only get results with s=.998 and lognorm(s), but s is restricted to be in (-1,1), when my skew parameter is -2.33.

Community
  • 1
  • 1
Jared
  • 3,651
  • 11
  • 39
  • 64
  • The docs list an explicit formula, http://docs.scipy.org/doc/scipy-0.17.0/reference/generated/scipy.stats.lognorm.html. what is your question again? – ev-br May 17 '16 at 16:53
  • All it says is it requires `s` and gives an example for `s=.954`. What is the `s`, and what are the restrictions on it? For my data I have a skew number of -2.33. – Jared May 17 '16 at 18:49
  • 1
    I want to draw random variables from a lognormal distribution with skew λ, mean μ, and standard deviation σ. – Jared May 17 '16 at 18:50

0 Answers0