Hacker News Reader

From the iPython notebook that appears to be associated with the blog post [1], they appear to be using Stan [2]. Stan is a relatively mature PPL that was designed with specific priorities in mind such as treating continuous variables. Overall, it's quite cool to see a startup focused on developing a PPL in the wild!

[1] https://github.com/fastforwardlabs/anomaly_detection/blob/ma...

[2] http://mc-stan.org

*edit, formatting to get line-breaks.

I recently developed a probabilistic programming framework in C#. There's a fun example [1] in the tutorial that should give you a pretty good idea of what a PPL "feels like" to use.

The example is originally from the Anglican website (another PPL), but the general idea is transferable; PPLs make composing entire probabilistic models as easy as adding integers. We can perform complex inference on our models, and even compose our inference methods to form more powerful ones.

[1] https://github.com/joashc/csharp-probability-monad#modelling...

The notebook linked to by the blog post uses Stan (and PyStan).