fancidev

Thanks for the update @tupui. A small suggestion for the `rv_continuous.fit` method for your consideration in the new framework: the `method` parameter should have default value `None`, with the below...

Another really small (but breaking) suggestion: the first parameter of `ppf` and `isf` of `rv_continuous` should read `p` not `q`, because it’s a probability rather than a quantile. (This change...

I tested the chosen bounds for different values of $r$ between $0$ and $1$, with more values near 0 and 1. Test script ```python import numpy as np import scipy.special...

> Thanks for the thorough analysis @fancidev ! This is a big improvement over what we had before. 2 points left: > > * Could you add tests for the...

@dschmitz89 I added test cases for the new branches. As for adding “Amos (1973)” to the reference page, I feel it doesn’t fit the topic there as the paper is...

> I noticed some issues still when comparing against the generic fit method, will update later. Cannot find how to dismiss my old review (Github ..). Thanks for checking! Maybe...

Thanks for the detailed comparison @dschmitz89 ! The difference in fitted log likelihood value between `fit()` and `fit(superfit=True)` shows a similar histogram (like the one you posted above) using the...

Thanks for the analysis @dschmitz89 ! I rebased the changes and they should be able to merge now. I'll look into the remaining numerical noise issue afterwards. Thanks again for...

> Can I interest you in doing a similar PR for the Von Mises-Fisher distrubution [xref](https://github.com/scipy/scipy/pull/20102#issuecomment-1951059513)? Otherwise, I will look into it in the coming weeks. Sure. I’ll look into...

The relationship between `genpareto`, `pareto`, and `powerlaw` is as follows. A `genpareto` distribution with shape $c \ne 0$, location $m \in \mathbb{R}$, and scale $s > 0$ has density $$...