add t distribution

[dswah]

would be nice to have for robust modelling

[dswah]

or a laplace distribution for seeking the median.

[dswah]

we can regress arbitrary quantiled using the asymmetric laplace distribution:

https://academic.oup.com/biostatistics/article/8/1/140/252234 https://en.m.wikipedia.org/wiki/Asymmetric_Laplace_distribution

[dswah]

turns out that this is probably NOT a good first issue :)

unfortunately, the work involved for adding a non-exponential family distribution is actually quite involved.

https://www.tandfonline.com/doi/full/10.1080/01621459.2016.1180986

[arose13]

It's crazy how this is so easy using Bayesian Inference and so hard to formulate a working version in pyGAM. HOWEVER, I see if anything can be done about this