85
$\begingroup$

A discussion in fall 2010 considered the extent to which purely software-related questions would be welcome on this site. It didn't really reach a conclusion, but one useful suggestion that arose is to collect a set of links to online support resources (such as user groups and list servers) for the various statistical computing platforms. I guess that would let us close some of these questions in a constructive and relatively guilt-free manner. I, for one, would like to help the people who come by with questions about SAS macro syntax or table access, even though these questions have no direct statistical interest (and would interest only small subcommunities here) and therefore ought to be closed or migrated.

Could we organize replies in the present thread by software platform? The ones of most immediate use are those that keep showing up: R, SAS, SPSS, Stata, Excel.

$\endgroup$
1
  • 6
    $\begingroup$ The (main) FAQ now links to this thread. $\endgroup$
    – whuber Mod
    Apr 7, 2011 at 6:51

24 Answers 24

Reset to default
75
$\begingroup$

R

R-help, and the various R Mailing Lists or SIGs, welcome any questions (provided they conform to the Posting Guide). Answers generally come within one or two days.

Quick-R gives a gentle overview of most of the basic R syntax for people coming from SAS, SPSS, or Stata. Stack Overflow also provides strong support for R questions. Additionally, rseek.org provides a custom Google search that facilitates queries related to R code, packages, articles, etc. You can search the R documentation and package documentation using rdocumentation.org site. Crantastic.org features useful reviews of current packages.

The UCLA Academic Technology Servicesprovide many worked examples of statistical analyses in R.

If you're looking for visualization ideas, visit the R Graph gallery and the Learn R blog, both of which feature a wide variety of plots and the accompanying code. The R Graphical Manual also provides a visualization of all CRAN R package example plots, and is searchable by topic. The Cookbook for R page provides multiple examples and recipes for plotting data (mostly using ggplot2) plus some additional information on using R.

To add the excellent resources list above, I (@michelle) have found the R Tutorial web site to be helpful. Also I have R bloggers as a feed. That has lots of useful posts from various bloggers and is an excellent way to keep up with new packages and new ways of using existing packages. If you're coming from SAS or SPSS, check out R for SAS and SPSS Users; there is a book with more information in it. An equivalent book for Stata users coming to R is R for Stata Users. Nice introductory tutorial can be found on R Tutorials page - it covers introduction to basics of R, using statistical tools such as t-tests, ANOVA, regression and other topics.

There are also some resources listed on our site here: Free resources for learning R, and on our R tag wiki.

For learning on more advanced topics in R programming the best resource that is available online is Advanced R site by Hadley Wickham. It is an online version of book under the same title. Another resource covering programming issues is The R Inferno by Patrick Burns available as pdf file. Those two cover topics that are negligible to most people that use R for statistics but can be crucial if you do actual programming in R and can be helpful in understanding how R works 'under the hood'. If trying to understand better how some R function works, you can always check their source code as R is open-source.

$\endgroup$
0
42
$\begingroup$

Python

Although this is not a statistical package per se, it has extensive statistical capabilities.

  • Matplotlib is a library for visualization. It has an assortment of tutorials and a gallery of examples in its documentation.

  • The Python Wiki is a gateway to Python.

  • Python Forum is a Python-oriented Q&A site.

  • From Python 3.4, the "statistics" module will be added to the standard library.

  • Pingouin is a pandas-aware library which provides diagnostic plots, power analysis, and various statistical tests.

  • StatPy ("special emphasis on astrostatistics").

  • Links to numerical and scientific packages.

  • Scikit-learn for machine learning.

  • SciPy.stats contains numerous classes for probability distributions and statistical tests. Other submodules of SciPy also provide numerical integration and integral transforms.

  • SemoPy2 provides trainable model classes for structural equation models. They support exploratory and confirmatory approaches, random effects, latent factor scores, mean components, among other features.

  • statsmodels, for a variety of models including generalized linear models, Markov models, and autoregressive models. Support for M-estimation, generalized least squares, maximum-likelihood, and seemingly-unrelated regression are included.

  • SymPy.stats provides a basic computer algebra system for symbolic statistical computing. The main module also provides a variety of mathematical tools such as derivatives, integrals (including Fourier transforms), and other tools that are useful in mathematical statistics.

  • pandas, for enhanced data structures, including time-series.

In addition, people who prefer Python for scripting, but would like access to R's wide-ranging statistical capabilities can call R from Python with rpy2 (see also: A Slug's Guide to Python).

$\endgroup$
2
  • 13
    $\begingroup$ NumPy and related questions get jumped on pretty quickly on Stack Overflow ... As a python programmer, I can actually recommend Stack Overflow as a default. $\endgroup$
    – Tritium21
    Sep 25, 2015 at 4:50
  • $\begingroup$ Tensorflow, Tensorflow Probability ( and the pytorch equivalents) for scalable probabilistic programming $\endgroup$
    – Georg M. Goerg
    Nov 27, 2022 at 17:13
31
$\begingroup$

Stata

Statalist is the place to go with questions about how to do things in any version of Stata. It is very active: Stata developers from StataCorp and many experienced users are leading members. Questions cover basic Stata use, Stata programming, and statistical practice. Before posting do study the FAQ Advice here. People are asked to use their real names. As a resource with very many answered questions and for its collective expertise, Statalist is ahead of all other forums.

Questions about Stata syntax, programming and output are addressed on StackOverflow. It's best to note its aim of being a forum for professional and enthusiast programmers and its focus on specific problems with people's own code. That is, SO is not a good forum for questions that are entirely or mostly statistical, including why or how Stata commands work.

Reddit occasionally includes Stata questions. Beginner questions are common. The number of experienced users lurking there is far smaller than on Statalist and advice is variable. There is more tolerance there for students seeking advice or assistance with their assignments, a plus or minus depending on your status and your question.

Much the same applies to Talk Stats, which appears to have faded away as a forum for Stata questions. EDIT 12 May 2023 This site now appears defunct.

Both of these last two sites allow anonymous identifiers, a plus if that is your preference.

Stata Forum.De is dedicated to Stata and is conducted in German. There is now little traffic.

Quora supports posting of opinions about Stata and its alternatives. Specific threads based on coding seem rare at best.

There is a Stata Users Group on Facebook. It supports questions in Spanish as well as in English.

Stata is often mentioned on Twitter but only a few posts ask for or provide support on specific questions.

The official FAQs are extremely useful as well. The help files for specific commands are available online; for direct access, form the url as http://www.stata.com/help.cgi? appended with the command name, like regress: http://www.stata.com/help.cgi?regress. The full pdf documentation is also available online, e.g., the User Guide and the Reference manuals are usually a good place to start for general commands.

StataCorp also maintains a YouTube training channel and offers registered users free technical support via e-mail.

ATS/UCLA has an entire section dedicated to Stata. Start from there with the learning modules, the FAQs or the "links by topic."

The StataCorp website carries information on, and a way to purchase, books on Stata: see this page.

$\endgroup$
4
  • 5
    $\begingroup$ I'm surprised no one mentioned Stata's official command documentation, which can be found online. Normally a google search for the command will bring it up in the first few results. To narrow the search, you can add "site:www.stata.com/help.cgi" after searching for the name of the command, e.g. regress site:www.stata.com/help.cgi. $\endgroup$
    – Ricardo Altamirano
    Nov 7, 2012 at 19:40
  • 2
    $\begingroup$ The pdf documentation that comes bundled with Stata is also quite excellent and well-written. Some of these manuals can even be found online. Lots of examples, formulas, and syntax. For some reason, many people seem to be unaware of their existence, even when they have used Stata for years. The link at the very top of Stata's documentation that you get when you type -help XXX- takes you there. $\endgroup$
    – dimitriy
    Apr 13, 2013 at 3:48
  • 8
    $\begingroup$ As from Stata 13, the entire manuals are on-line and accessible to all. $\endgroup$
    – Nick Cox
    Oct 25, 2013 at 23:59
  • $\begingroup$ Typing -help XXX- into an internet search engine usually spits out the Stata help file as the first or second search result (at least for Google and Yahoo, anyway) $\endgroup$
    – Marquis de Carabas
    Apr 13, 2016 at 5:08
24
$\begingroup$

MATLAB

MATLAB (MATrix LABoratory) is a multi-paradigm numerical computing environment and fourth-generation programming language. It is developed by MathWorks. MATLAB has a free, open-source counterpart named Octave that is distributed on GNU-GPL license and offers access to a subset of MATLAB's original functionality.

MATLAB questions get routinely answered in SO. In addition to that one can check:

$\endgroup$
23
$\begingroup$

Julia

Julia is a new language with MATLAB-like syntax but Lisp-like semantics and a Lisp-style macro language. Julia has growing capabilities for statistics, and, its main advantage, is blazing fast! To learn about Julia start with http://julialang.org/ and, especially, http://julialang.org/community/.

Julia has several online discussion groups including julia-users and julia-stats.

$\endgroup$
21
$\begingroup$

SAS

SAS questions do get asked and answered on StackOverflow; SAS also runs a community forum, which is very active.

The proceedings of SUGI are a tremendous resource. Also invaluable if you need material for a corny SAS stand-up comedy routine.

The UCLA Academic Technology Services provide really fantastic resources for SAS.

Another resource is SAS-L (site has archives and information on how to join).

The online documentation at the SAS support page is a great resource. It includes the SAS User's Guide which contains quite detailed information on SAS procedures, including syntax, theoretical details, and examples. As an example, here's the page for proc glm.

Lex Jansen's site indexes not only all of SAS Global Forum (formerly SUGI) but the regional meetings as well.

$\endgroup$
1
  • $\begingroup$ The link to 'various SAS blogs' via Chris Hemedinger's blog is also no longer usable. Google reader no longer exists, so there's no way of getting to the list. $\endgroup$
    – Rob Penridge
    Nov 12, 2015 at 16:48
18
$\begingroup$

Stan

Stan is an open-source, probabilistic programming language implementing full Bayesian statistical inference and penalized maximum likelihood estimation.

Stan has very good documentation (including detailed Language Manual), github Wiki page and active Stan users mailing list.

$\endgroup$
16
$\begingroup$

SPSS

Several forums devoted to SPSS software usage are;

SPSSX and the google group forum receive a fairly wide variety of data manipulation questions and questions related to statistical analysis. Developer central is sometimes a more appropriate place to go if you are doing anything not related to actions that can be accomplished through traditional syntax or the GUI (e.g. scripting in python or vba).

Make sure to check out the SPSS tag wiki for a more complete list of resources, many of which are freely accessible online.

$\endgroup$
14
$\begingroup$

JAGS

JAGS is Just Another Gibbs Sampler. It is a program for analysis of Bayesian hierarchical models using Markov Chain Monte Carlo (MCMC) simulation not wholly unlike BUGS. JAGS was written with three aims in mind:

  • To have a cross-platform engine for the BUGS language
  • To be extensible, allowing users to write their own functions, distributions and samplers.
  • To be a plaftorm for experimentation with ideas in Bayesian modelling

(source: http://mcmc-jags.sourceforge.net/)

Online forum for JAGS users can be found on JAGS sourceforge page.

$\endgroup$
13
$\begingroup$

RapidMiner

$\endgroup$
11
$\begingroup$

Minitab

Learn more about Minitab on Wikipedia

$\endgroup$
10
$\begingroup$

JMP

JMP is a desktop statistical exploration tool from SAS.

The primary source of online support for JMP is the JMP User Community site which hosts discussion forums and a file exchange for add-ins and data sets.

Other online resources include JMP documentation, a semi-technical JMP blog and weekly live webcasts.

$\endgroup$
1
  • $\begingroup$ Too bad this isn't nested. JMP script (JSL) is very particular and demanding. It might qualify as its own sub-heading. $\endgroup$
    – EngrStudent
    Jan 20, 2015 at 21:08
10
$\begingroup$

Fortran

Fortran is one of the main languages in which statistical algorithms have been coded.

Comp.lang.fortran is an active Usenet group.

Stack Overflow often has Fortran questions.

The Open Directory has links to many Fortran codes in statistics and econometrics.

$\endgroup$
9
$\begingroup$

Wizard

Wizard is desktop statistics and data visualization package for Mac OS X. The primary support channel is the Wizard User Group hosted on Google Groups.

Disclosure: I am the developer of the software.

$\endgroup$
0
8
$\begingroup$

Weka

Weka is a collection of machine learning algorithms for data mining tasks. The algorithms can either be applied directly to a dataset or called from your own Java code. Weka contains tools for data pre-processing, classification, regression, clustering, association rules, and visualization. It is also well-suited for developing new machine learning schemes.

A mailing list, forum, & IRC channel are listed at the Weka site under Getting Help

$\endgroup$
6
$\begingroup$

Taverna Workbench

It is a open source tool that has been more used for helping in data mining recently. In other words, you can create pieces of the workflow to code extracting data from different databases, analyze them, do some processing and it also supports output in R.

"Taverna is an open source and domain-independent Workflow Management System – a suite of tools used to design and execute scientific workflows and aid in silica experimentation".

$\endgroup$
6
$\begingroup$

TensorFlow

TensorFlow is an open-source library for machine learning that was developed by Google. It has detailed documentation on its web site. Bugs and feature requests can be posted and viewed on TensorFlow's github page.

Introductory tutorials on TensorFlow can be found on Udacity and Coursera online courses on deep learning.

$\endgroup$
1
  • 1
    $\begingroup$ On the TensorFlow discussion group, it is stated that the group is not for seeking help and advice. They suggest to go to StackOverflow. $\endgroup$
    – Patrick Hund
    Jul 9, 2018 at 7:36
5
$\begingroup$

NONMEM

A fortran based non-linear mixed effects modelling software with very powerful algorithms including ODE solvers. It is a commercial software made by ICON plc that is widely used in the pharmacometrics community. It has a steep learning curve but once you get used to it, you'll probably not need to use another non-linear mixed effects modelling software. It needs to be used alongside other statistical tools like R to analyse and visualise modelling results. It can also be used alongside PsN for automated covariate analysis and visual predictive checks, simulation, etc. For more on NONMEM, visit ICON plc.

There is a NONMEM user group dedicated to users of NONMEM here.

This short tutorial gives a very concise introduction to pharmacometrics and non-linear mixed effects modelling.

$\endgroup$
1
  • 2
    $\begingroup$ Ahh.. Pharmokinetics people.. bless them. They did NLME models before it was cool. $\endgroup$
    – usεr11852
    Aug 10, 2016 at 7:54
5
$\begingroup$

ROOT

ROOT is a general purpose data-analysis framework written in C++. It is the de facto choice for any kind of analysis in the particle physics community, although it is not limited to that community. It provides many specialised functionality through libraries, e.g.:

  • Minuit is a minimisation library original written in Fortran, now reimplemented in C++,
  • RooFit is a fitting framework focused on maximum likelihood fitting,
  • RooStats is a statistics package built around RooFit used to provide tools for statistical significance calculations, limit setting, etc,
  • TMVA is a multivariate analysis library which implements numerous machine learning algorithms like neural networks, decision trees, etc.

ROOT provides language bindings in other programming languages like, Python, Go, and Ruby. It also provides various I/O facilites to handle very large datasets (multi-gigabyte), as well as visualisation facilities.

The package is developed as a joint effort by Fermilab and CERN. The homepage has extensive documentation that include getting started tutorials, code examples, as well as an extensive reference manual. To get help, one can go to the web forum or the mailing list. Unfortunately to signup for the mailing list, one needs to get a lightweight CERN account (no restrictions though, anyone can get it).

$\endgroup$
4
$\begingroup$

Mplus

Mplus is a latent variable modeling program. It offers a wide range of models, estimators, and algorithms to researchers. Its analysis capabilities include a variety of basic and advanced analyses, with a special focus on analyses that include the estimation and testing of latent variables (e.g., Structural equation modeling).

For more information, visit the Mplus website which provides tutorials, user guides, and other sources. Additionally, Mplus Discussion provides a platform for users to share questions and comments about the program and modeling issues.

$\endgroup$
3
$\begingroup$

PyTorch

PyTorch comminity hosts a list of educational resources on their page, including Discuss forum and Slack channel.

$\endgroup$
3
$\begingroup$

Keras

Keras is a high-level neural networks API, written in Python and capable of running on top of TensorFlow, CNTK, or Theano. It was developed with a focus on enabling fast experimentation. Being able to go from idea to result with the least possible delay is key to doing good research.

It has detailed online documentation, users can seek help in Keras Google group and the Slack channel. Moreover, StackOverflow and Data Science Exchange both have a tag for Keras.

$\endgroup$
2
$\begingroup$

Vowpal Wabbit

VW is a command-line machine learning software that is designed to efficiently work with large amounts of data, that is not limited by RAM. It supports online learning. The software gained popularity as a tool used by multiple users on Kaggle competitions.

As described on its github page:

Vowpal Wabbit is a machine learning system which pushes the frontier of machine learning with techniques such as online, hashing, allreduce, reductions, learning2search, active, and interactive learning. http://hunch.net/~vw/

Pretty detailed documentation with code examples can be found on the github Wiki page of the project. Video lecutres by John Langford, creator of VW, giving introductory tutorials can be found on YouTube here and here. There is also arXived paper "A Reliable Effective Terascale Linear Learning System" by Agarwal et al describing theoretical basis of the software. Users can seek support on VW mailing list.

$\endgroup$
1
$\begingroup$

ACER ConQuest

ACER ConQuest is a computer program for fitting both unidimensional and multidimensional item response and latent regression models. It provides data analysis based on a comprehensive and flexible range of item response models, allowing examination of the properties of performance assessments, traditional assessments and rating scales.

Support is provided via:

Download a demonstration copy

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .