DOC: correction / update of docstrings of distributions in scipy.stats #9170
Conversation
chrisb83
added
scipy.stats
Documentation
scipy/stats/_continuous_distns.py
Outdated
| where ``lambda = (K * scale)**(-1)``. | ||
| It can be thought of as the sum of a standard normal random | ||
| variable and an exponentially distributed random variable with rate | ||
| ``1/K``. |
There was a problem hiding this comment.
In wikipedia and previous definition loc and scale are only used for the normal part.
How does this match up with the loc and scale shift that we have generically.
i.e. there should be a some explanation about the effect of scale, my guess similar to the old definition of lambda
or how do we need to choose k if we have lambda and normal-scale?
There was a problem hiding this comment.
In standard form, with X standard normal and Y exponential with rate 1/K , X + Y is exponnorm, thus loc + scale*(X + Y) is the sum of norm(loc, scale) and scale * Y and the rate parameter of the exponential distribution scales as stated in the old documentation.
I think it is better to state the standard form only this point, the scaling is explained after the notes with a reference to wikipedia.
|
LGTM, thanks @chrisb83 , @josef-pkt |
I tried to correct minor issues in the docstrings of continuous distributions.
Closes #6358
Closes #8878 (clarify difference between vonmises and vonmises_line)