The result was no consensus. — Coffee // have a cup // beans // 04:21, 25 March 2017 (UTC)
Non-notable. The only reference covering it in any depth was written by the creators of the software. No evidence that any third party has taken such an interest in it. Anyone can upload some math routines to Github and write about them. It needs much more than this for a software package to be notable. JohnBlackburnewordsdeeds 19:51, 22 February 2017 (UTC)
I can add more references, if that is needed. Some of them could be:
There is a similar page for a similar software https://en.wikipedia.org/wiki/Stan_(software) — Preceding unsigned comment added by Aloctavodia (talk • contribs) 20:03, 22 February 2017 (UTC)
There is also other software based on PyMC3.
Bambi: BAyesian Model-Building Interface (BAMBI) in Python. NiPyMC: Bayesian mixed-effects modeling of fMRI data in Python. gelato: Bayesian Neural Networks with PyMC3 and Lasagne. beat: Bayesian Earthquake Analysis Tool. Edward: A library for probabilistic modeling, inference, and criticism. — Preceding unsigned comment added by Aloctavodia (talk • contribs) 20:15, 22 February 2017 (UTC)
> No evidence that any third party has taken such an interest in it.
Speaking as s statistician who did not write pymc3 and yet uses it, this particular assertion seems questionable - there are many blog posts using it, more than 200 stackoverflow questions about it, hundreds of academic articles written using it, heavy and sustained source code contribution and so on. If it were a university professor, its publication record would be likely above-the-median.
It is, however, used for machine-learning and statistics, which might contribute to the perception that it is obscure amongst people outside those fields. --Livingthingdan (talk) 23:16, 23 February 2017 (UTC)
I have added new sources to the article. — Preceding unsigned comment added by Aloctavodia (talk • contribs) 17:05, 11 March 2017 (UTC)