Design your data analysis process through visual programming. Orange remembers your choices, suggests most frequently used combinations, and intelligently chooses which communication channels between widgets to use.

Orange is packed with different visualizations, from scatterplots, bar charts, trees, to dendrograms, networks and heatmaps.

Actions seamlessly propagate through data analysis schema. Selection of data subset in one widget can automatically trigger change of display in the other one. By combining various widgets you can design data analytics framework of choice.

Over 100 widgets and growing. Coverage of most of standard data analysis tasks. Also specialized add-ons are available, like Bioorange for bioinformatics.

With scripting interface in Python, programming new algorithms and developing complex data analysis procedures is pure joy, using and reusing all power found in v

REST for the sneakernet - how to send your next few likely uses for a sporadically available API by sending a queued decision tree of API calls.

high performance bayesian inference over complex graphical models.

FACTORIE is a toolkit for deployable probabilistic modeling, implemented as a software library in Scala. It provides its users with a succinct language for creating relational factor graphs, estimating parameters and performing inference.
It is object-oriented... in the definition of random variables, factors, inference and learning methods.
It is scalable, with demonstrated success on problems with many millions of variables and factors, and on models that have changing structure, such as case factor diagrams... capable of handling billions of variables.
It is flexible, supporting multiple modeling and inference paradigms. Its original emphasis was on conditional random fields, undirected graphical models, MCMC inference, online training, and discriminative parameter estimation... has preliminary support for variational inference, including belief propagation and mean-field methods.

combo data-mining/NLP/web-scraping toolkit for instant natural experiments online

you don't have to leave python to R for nearly as much these days.

This wiki has information on solvers and problems that arise in these fields (and subfields, such as compressed sensing).
Everyone is welcome and encouraged to edit this wiki; it runs
Contents

The contents of this wiki have been organized into categories.
Problems formulations that arise in sparse- and low-rank approximation.
Solvers that are used for solving these problems. There are many sub-categories of solvers, such as:
Convex solvers
Greedy solvers
Matrix completion solvers
and many more (all of them listed at Solvers ).
Benchmarking/Test problems for comparing algorithms
Applications (in software) of sparsity or low-rank based techniques
Hardware devices that perform compressed sensing.
...

How to do stats on the surface of he earth with beautiful visualizations

python minority game simulation

a growth industry: automating the production of journalism, books, textual content

weirldy relaxed academic dispensing soothing quotes:
Do not burn yourselves out. Be as I am - a reluctant enthusiast, a part-time crusader, a half-hearted fanatic. Save the other half of yourselves and your lives for pleasure and adventure. It is not enough to fight for natural land and the west; it is even more important to enjoy it. While you can. While it's still there... Enjoy yourselves, keep your brain in your head and your head firmly attached to the body, the body active and alive, and I promise you this much: I promise you this one sweet victory over our enemies, over those desk-bound men with their hearts in a safe deposit box, and their eyes hypnotized by desk calculators. I promise you this: you will outlive the bastards. --Ed Abbey.

If dsge performs better than the competitors, does that mean that it performs still?

A gentle climb up a staircase, atop which is ARIMA.

"So, this post is a bit of a brain dump on rich data structures in Python and what needs to happen in the very near future. I care about them for statistical computing (I want to build a statistical computing environment that trounces R) and financial data analysis (all evidence leads me to believe that Python is the best all-around tool for the finance space). Other people in the scientific Python community want them for numerous other applications: geophysics, neuroscience, etc. It’s really hard to make everyone happy with a single solution. But the current state of affairs has me rather anxious. And I’d like to explain why..."

"pandas is a python package providing convenient data structures for time series, cross-sectional, or any other form of “labeled” data, with tools for building statistical and econometric models."

handle data in python intuitively. pass to R for fiddly bits.

neat alternate perspective on statistics, in that kind of living-in-a-funding-bubble US NIST kind of way. Bit of an exploratory daya analysis focus, but still worthwhile.

a great collection of mathquirk, all online and free

The uniform uncertainty principle and compressed sensing

Best explanation of this sparse image representation thing that I have yet seen. Well wicked.

categorical data has other Shannon-information-like estiamtors of "interestingness"

native python graph handling with ultralight api built aroudn hashes.

c graph library with r and python support

c++ graph lib for python, optimised for performance.

Prediction as a game against nature. Weird science.

depicting PCA versus ICA using scipy

TANEJA. I.J. (2001), - this has the stuff about norms and maximum entropy measures etc and their implications

"In that sense, when people play with loss functions, they are essentially also playing with probability distributions that are entailed by the loss functions. When they use L1 or L2 regularization for regression, they are picking either Gaussian or Laplace priors for the parameters. The reason for the popularity has been primarily the realization that Laplace prior is better than Gaussian prior on many benchmarks. I wonder if the log(1+d^2) norms will generate as many papers as L1, or whether statisticians will migrate from Student to Laplace."

sweet looking machine learning toolchain for python

hawt python statistics. If you don't actually like R THAT much.

statiscial analysis on your database contents: "MADlib is an open-source library for scalable in-database analytics. It provides data-parallel implementations of mathematical, statistical and machine learning methods for structured and unstructured data". Seems to be largely in PL/C, with some bonus python on the front.

interesting bunch of courses on statistical data fiddling

science communication expert explains probability with rare precision. excellent for tips about communicating risk and lots of tasty examples (comparing hose riding with ecstasy, motorbikes with pregnancy, all that stuff)

handy performance tip example -wrap gsl library calls in cython

optimising crazy inner loops in cython using GSL calls.

Rubt's sparse but serviceable answer to numpy

not pretty, but necessary. How to comunicate statistics and risks to people in a matter which suffers least at the hands of our inherent cognitive biases

interesting little wiki of complexity and emergence for the student-type.

the case for replacing the language of R with somethig else that can access the same statistical power

how to find, in a statistically valid fashion, that your data fits a power law.

best explanation of mutual_information on the interweb,

MCMC, Monte Carlo methods, and the Metropolis -Hastings sampling doohickey for fun and profit. Well, profit.

Bayesian estimation, particularly using Markov chain Monte Carlo (MCMC), is an increasingly relevant approach to statistical estimation. However, few statistical software packages implement MCMC samplers, and they are non-trivial to code by hand. PyMC is a python module that implements the Metropolis-Hastings algorithm as a python class, and is extremely flexible and applicable to a large suite of problems. PyMC includes methods for summarizing output, plotting, goodness-of-fit and convergence diagnostics

numerical recipes in c, 2nd edition, is online.

the statistics and evolution of interesting random cellular automata

MI and H estimates

information theoretic widgets and doohickeys for python statistics

the python<->R interface is being rebooted

Tufte-compliant graphing tricks for R

handy image processing code in python, including similarity detection, mutual information volumetric rendering and so on

it's the page of that trendy work-out-your-laws-for-you package

This guy is reliably entertaining. in this case, it's a userful, arrogant rant about what you need to know about stats to push beyond the "mean request time or whatever. highlights:

>The next day we had IBM fixing the problem (turned out to be a single update index command) and we all kept our jobs. That’s what a proper analysis method can do for you.

>still I see software developers begging for gazillions of dollars to buy some crap tool that doesn’t even mention “standard deviation”, but throws “user” around like it’s Dr. Phil treating Robert Downey Jr. for heroin addiction.

online free ebook about the stats and tools needed for all that trendy geospatial shit to happen in a rigorous way

network stats for R

some nice hacks for validating agent based models

parser for probabilistic context free grammars using the CYK algorithm found in the Stanford NLP parser, but faster and python-happy

part I of the SFI language deep history project

part II of the SFI language deep history project

new compression algorithms which works on streams and is smaller than most stuff. licence unclear. background theory intriguing.

learning in the game-theoretic statistical framework of vovk and shafer

Larry Wasserman's own course page for stats 2

I always fall down on the step about not trying to be clever.

monte carlo methods for dummies. Pellucid.

inferring networks given connection weights, and generating them.

Judea Pearl on the problems of using probabilistic reasoning when the human mind tends to causal

tasty howto maxent guides from UC Davis

Shalizi's package for inferring hidden recursive markov model values and strucutre from a time series

"Peach is a pure-python module, based on SciPy and NumPy to implement algorithms for computational intelligence and machine learning. Methods implemented include, but are not limited to, artificial neural networks, fuzzy logic, genetic algorithms, swarm intelligence and much more."

including graph theoretic tools.

the implications of statistical approaches to intelligence gathering. we're all part of the mass society now, eh?

magickal classifiers and predictors on arbitrary data, courtesy google. probably handy for stuff, esp if they actually disclosed the algorithm.

a whole bunch of stats manuals online.

nice statmech/thermal physics textook

nice statmech/thermo mech textbook

export from netlogo to R

learn r monkey-see-monkey-do style