Add distributions

[aloctavodia]

This is a list to keep track of the currently implemented distributions and the ones still missing.

• asymmetric_laplace
• bernoulli
• betabinomial
• beta
• betascaled
• binomial
• categorical
• cauchy
• chi_squared
• discrete_uniform
• discrete_weibull
• exgaussian
• exponential
• gamma
• geometric
• gumbel
• halfcauchy
• halfnormal
• halfstudentt
• hypergeometric
• inversegamma
• kumaraswamy
• laplace
• logistic
• logitnormal
• loglogistic
• lognormal
• moyal
• negativebinomial
• normal
• pareto
• poisson
• rice
• scaled_inverse_chi_squared
• skewnormal
• skew_studentt
• studentt
• triangular
• truncatednormal
• uniform
• vonmises
• wald
• weibull
• zi_binomial
• zi_negativebinomial
• zi_poisson
• Multivariate Normal
• Dirichlet

Modifiers (for the moment, we are going to let other libraries like pymc/preliz to handle these)

• censored
• hurdle
• mixture
• truncated

Implemented in PyMC, not PreliZ

• PolyaGamma
• DirichletMultinomial
• Multinomial
• MvStudentT
• MatrixNormal
• Wishart
• LKJCorr
• StickBreakingWeights
• ???

[aloctavodia]

We almost have all the distributions that are implemented in PreliZ!!! Except for some modifiers.
We also need to extend this list to include distributions implemented in PyMC. I don't think we want to implement here stuff like timeseries, orderedprobit, zerosumnormal, etc @fonnesbeck, @juanitorduz, @jessegrabowski

[fonnesbeck]

I can knock out a few more over the weekend.

[jessegrabowski]

time series might be interesting though, to have the stationary mean/covariance matrix of the process. I understand if it's out of scope

[aloctavodia]

We have to define the scope and what we think is useful to have here, and of course that could change in time depending on our needs.