datahaskell / docs Goto Github PK

https://hackage.haskell.org/package/hip

NumHask

Sparkle

• generation of random numbers
• statistical tests
• hypothesis testing

Packages:

• statistics

"How to produce numerically stable/reliable Haskell programs? "

• Numerically stable summation : Numeric.Sum in https://hackage.haskell.org/package/math-functions
  • case in point:

> foldr (+) 0 [1.0, 10.0e100, 1.0, (-10.0e100)]
1.0

> foldl (+) 0 [1.0, 10.0e100, 1.0, (-10.0e100)]
0.0

import Numeric.Sum -- math-functions 
> Numeric.Sum.sum kbn [1.0, 10.0e100, 1.0, (-10.0e100)] 2.0

http://www.datahaskell.org/docs/tutorial/ihaskell-installation.html

Frames

It seems there is a problem with the markdown engine.

Everything's fine on my local machine:

But this is how the markdown is rendered at datahaskell.org/docs:

According to this page we're using kramdown which should just work fine.

• Template embeds Hypothes.is annotation JS : https://h.readthedocs.io/projects/client/en/latest/publishers/embedding/
• Create DH Hypothes.is user group

The point of the benchmarks is to compare different numerical libraries, so they should be formatted in a way that allows a user to see how long common operations take using different libraries.

PlotlyHS

I think it would be great to keep our "Current environment" [1] directory in sync with the (generally better known) "State of the ecosystem" [2]

[1] http://www.datahaskell.org/docs/community/current-environment.html

[2] https://github.com/Gabriel439/post-rfc/blob/master/sotu.md

Hi, I'm curious to know why sparkle (https://github.com/tweag/sparkle) isn't beside kraps-h? I am not a data scientist, nor have I used Spark.

The roadmap page is just empty, with a copyright date 6 years ago, which is pretty confusing. It may give the impression that this is a dead project.

It's outdated as it uses the 0.0.1 version, and now Haskell.do is 0.9.5

Transient

I get the following build error in my mail:

The page build failed with the following error:

Page build failed. For more information, see https://help.github.com/articles/troubleshooting-github-pages-build-failures.

For information on troubleshooting Jekyll see:

  https://help.github.com/articles/troubleshooting-jekyll-builds

If you have any questions you can contact us by replying to this email.

I contacted Github to see if they have an idea on how to fix this.
I've read those 2 links but still don't know what happens or why it happens.

Mentioning @DataHaskell/everyone in case someone knows how to solve this.

Write a howto/tutorial comparing a few ways of doing linear regression in Haskell.

Frequentist

• least-squares linear regression (e.g. with hmatrix)
• LASSO linear regression (?) (needs cyclical coordinate descent)
  • theory : Chapter 3 of Hastie, Tibshirani, Friedman

Bayesian

• Closed-form posterior
  • theory : Chapter 3 of Bishop
• Sampling-based (i.e. no closed form posterior)
  • e.g. Markov chain MC (via http://hackage.haskell.org/package/declarative )
  • theory and algorithms : e.g. http://matt-wand.utsacademics.info/webStatSem/bayLinRegWbLec2x2.pdf