5 ms·

Worth noting Prophet is R/Python wrappers to some models with reasonable defaults, written in and fit by Stan, a probabilistic programming language, and Bayesia

Worth noting Prophet is R/Python wrappers to some models with reasonable defaults, written in and fit by Stan, a probabilistic programming language, and Bayesian estimation framework. Stan is amazing in that you can fit pretty much any model you can describe in an equation (given enough time and compute, of course)! More on Stan here: http://mc-stan.org/ http://mc-stan.org/

dragandj 10y ago
... and if you like Clojure, you might try Bayadera, which has its own engine running the analysis on the GPU. http://github.com/uncomplicate/bayadera http://github.com/uncomplicate/bayadera
- bpicolo 10y ago
  Readme has neither useful docs, nor any link to docs. =/
- mej10 10y ago
  This looks like it could be awesome but it has almost no information about what its purpose is or how to use it.
  - dragandj 10y ago
    You are right. The docs have been due to be written for many months now, and that is the main reason the library has not been released yet. On the other hand, the test folder contains many tests, among them full examples from many chapters from the book Doing Bayesian Dara Analysis, recommended above.
    - mej10 10y ago
      Oh I see. That's cool! I do want to check it out but will probably wait for its release. And I wait with bated breath.
- diab0lic 10y ago
  I'm pretty interested in this as I do most of my work on the JVM and I love trying this out on our stream processor at work. Cloned and tried to build it but I'm getting an error regarding uncomplicate:commons:0.3.0-SNAPSHOT being unavailable on clojars. Is that something you currently have installed to your local maven repo? I don't see it here: https://clojars.org/repo/uncomplicate/commons/ https://clojars.org/repo/uncomplicate/commons/ I can get it to build with 0.2.2 but it is missing the "releaseable?" function. In any case this looks awesome and I'll be keeping an eye on it / playing with it for some new projects. EDIT: I was able to get it building by cloning your commons library and running "lein install". :)
  - dragandj 10y ago
    Please also note that it currently requires OpenCL 2.0 compatibility, and is optimized for AMD GPUs. I plan to add CUDA support later this year.
    - diab0lic 10y ago
      Yeah I'm running into issues with this using a mid 2015 MacBook pro w/ Nvidia gpu. Looks like it only supports OpenCL 1.2.
  - dragandj 10y ago
    Also, there are a few presentations availavle at http://dragan.rocks http://dragan.rocks. just follow the blog posts.
treigerm 10y ago
Do you know of any good beginner tutorials for Stan or probabilistic programming in general? All the examples that I found seemed quite complex and I was a bit overwhelmed by all the math. Which might also be a sign that I should brush up my math skills. What kind of math/stats should I revise to be able to better understand probabilistic programming?
- projectorlochsa 10y ago
  Probabilistic graphical models. That's the foundation. The way you set up your model is by nodes and edges that specify the flow of influence (directed or undirected). Then it seems that there are general methods for inference and learning on any kind of graph one might pose. For simple graphs (and simple is something one might want when modelling) the methods should be fairly effective. Unfortunately, the biggest book on the subject that I know (Koller & Friedman) isn't accessible. Koller's course is also not that accessible.
- muraiki 10y ago
  Doing Bayesian Data Analysis, Second Edition: A Tutorial with R, JAGS, and Stan. It is very approachable and also has lots of practice problems. It's not a math-heavy book at all. Edit: I wouldn't recommend Probabilistic Programming and Bayesian Methods for Hackers. When I tried using it, I felt that too much was glossed over. The book that I recommend excels at conveying a strong intuition for how these various techniques work.
  - saosebastiao 10y ago
    I'd second this. DBDA made things click for me where a handful of other books failed.
  - j_s 10y ago
    Doing Bayesian Data Analysis https://amzn.com/dp/B00PYZ2VR6/ https://amzn.com/dp/B00PYZ2VR6/ ($65). Free is hard to beat, especially for someone just testing the waters.
- jritchie 10y ago
multani-hn 10y ago
Stan is nice but its GPL license is taboo in my corporate environment :( . I am puzzled how they managed to release Prophet under BSD with such a dependency.
- matthjensen 10y ago
  Stan has a BSD core. Prophet must avoid the GPLv3 interfaces.
  - thisisdave 10y ago
    It doesn't avoid the R interface, which is GPL'd (version 3).
    - j_s 10y ago
      Smells like more Facebook licensing drama, something new this time instead of the revocable patent license!
  - shoyer 10y ago
    The Python API (pystan) is also GPLv3.

dragandj 10y ago

... and if you like Clojure, you might try Bayadera, which has its own engine running the analysis on the GPU. http://github.com/uncomplicate/bayadera http://github.com/uncomplicate/bayadera

bpicolo 10y ago
Readme has neither useful docs, nor any link to docs. =/
mej10 10y ago
This looks like it could be awesome but it has almost no information about what its purpose is or how to use it.
- dragandj 10y ago
  You are right. The docs have been due to be written for many months now, and that is the main reason the library has not been released yet. On the other hand, the test folder contains many tests, among them full examples from many chapters from the book Doing Bayesian Dara Analysis, recommended above.
  - mej10 10y ago
    Oh I see. That's cool! I do want to check it out but will probably wait for its release. And I wait with bated breath.
diab0lic 10y ago
I'm pretty interested in this as I do most of my work on the JVM and I love trying this out on our stream processor at work. Cloned and tried to build it but I'm getting an error regarding uncomplicate:commons:0.3.0-SNAPSHOT being unavailable on clojars. Is that something you currently have installed to your local maven repo? I don't see it here: https://clojars.org/repo/uncomplicate/commons/ https://clojars.org/repo/uncomplicate/commons/ I can get it to build with 0.2.2 but it is missing the "releaseable?" function. In any case this looks awesome and I'll be keeping an eye on it / playing with it for some new projects. EDIT: I was able to get it building by cloning your commons library and running "lein install". :)
- dragandj 10y ago
  Please also note that it currently requires OpenCL 2.0 compatibility, and is optimized for AMD GPUs. I plan to add CUDA support later this year.
  - diab0lic 10y ago
    Yeah I'm running into issues with this using a mid 2015 MacBook pro w/ Nvidia gpu. Looks like it only supports OpenCL 1.2.
- dragandj 10y ago
  Also, there are a few presentations availavle at http://dragan.rocks http://dragan.rocks. just follow the blog posts.

treigerm 10y ago

Do you know of any good beginner tutorials for Stan or probabilistic programming in general? All the examples that I found seemed quite complex and I was a bit overwhelmed by all the math. Which might also be a sign that I should brush up my math skills. What kind of math/stats should I revise to be able to better understand probabilistic programming?

projectorlochsa 10y ago
Probabilistic graphical models. That's the foundation. The way you set up your model is by nodes and edges that specify the flow of influence (directed or undirected). Then it seems that there are general methods for inference and learning on any kind of graph one might pose. For simple graphs (and simple is something one might want when modelling) the methods should be fairly effective. Unfortunately, the biggest book on the subject that I know (Koller & Friedman) isn't accessible. Koller's course is also not that accessible.
muraiki 10y ago
Doing Bayesian Data Analysis, Second Edition: A Tutorial with R, JAGS, and Stan. It is very approachable and also has lots of practice problems. It's not a math-heavy book at all. Edit: I wouldn't recommend Probabilistic Programming and Bayesian Methods for Hackers. When I tried using it, I felt that too much was glossed over. The book that I recommend excels at conveying a strong intuition for how these various techniques work.
- saosebastiao 10y ago
  I'd second this. DBDA made things click for me where a handful of other books failed.
- j_s 10y ago
  Doing Bayesian Data Analysis https://amzn.com/dp/B00PYZ2VR6/ https://amzn.com/dp/B00PYZ2VR6/ ($65). Free is hard to beat, especially for someone just testing the waters.
jritchie 10y ago

multani-hn 10y ago

Stan is nice but its GPL license is taboo in my corporate environment :( . I am puzzled how they managed to release Prophet under BSD with such a dependency.

matthjensen 10y ago
Stan has a BSD core. Prophet must avoid the GPLv3 interfaces.
- thisisdave 10y ago
  It doesn't avoid the R interface, which is GPL'd (version 3).
  - j_s 10y ago
    Smells like more Facebook licensing drama, something new this time instead of the revocable patent license!
- shoyer 10y ago
  The Python API (pystan) is also GPLv3.