R language on the GPU

Here are some nice articles on using R on Graphical Processing Units (GPU) mainly made by NVidia. Think of a GPU as a customized desktop with specialized computing equivalent to much faster computing. i.e. Matlab users can read the webinars here http://www.nvidia.com/object/webinar.html

Now a slightly better definition of GPU computing is from http://www.nvidia.com/object/GPU_Computing.html

GPU computing is the use of a GPU (graphics processing unit) to do general purpose scientific and engineering computing.
The model for GPU computing is to use a CPU and GPU together in a heterogeneous computing model. The sequential part of the application runs on the CPU and the computationally-intensive part runs on the GPU. From the user’s perspective, the application just runs faster because it is using the high-performance of the GPU to boost performance.

rgpu

Citation:

http://brainarray.mbni.med.umich.edu/brainarray/rgpgpu/

R is the most popular open source statistical environment in the biomedical research community. However, most of the popular R function implementations involve no parallelism and they can only be executed as separate instances on multicore or cluster hardware for large data-parallel analysis tasks. The arrival of modern graphics processing units (GPUs) with user friendly programming tools, such as nVidia’s CUDA toolkit (http://www.nvidia.com/cuda), provides a possibility of increasing the computational efficiency of many common tasks by more than one order of magnitude (http://gpgpu.org/). However, most R users are not trained to program a GPU, a key obstacle for the widespread adoption of GPUs in biomedical research.

The research project at the page mentioned above has developed special packages for the above need- R on a GPU.

he initial package is hosted by CRAN as gputools a sorce package for UNIX and Linux systems. Be sure to set the environment variable CUDA_HOME to the root of your CUDA toolkit installation. Then install the package in the usual R manner. The installation process will automatically make use of nVidia’s nvcc compiler and CUBLAS shared library.

and some figures

Figure 1 provides performance comparisons between original R functions assuming a four thread data parallel solution on Intel Core i7 920 and our GPU enabled R functions for a GTX 295 GPU. The speedup test consisted of testing each of three algorithms with five randomly generated data sets. The Granger causality algorithm was tested with a lag of 2 for 200, 400, 600, 800, and 1000 random variables with 10 observations each. Complete hierarchical clustering was tested with 1000, 2000, 4000, 6000, and 8000 points. Calculation of Kendall’s correlation coefficient was tested with 20, 30, 40, 50, and 60 random variables with 10000 observations each

Ajay- For hard core data mining people ,customized GPU’s for accelerated analytics and data mining sounds like fun and common sense. Are there other packages for customization on a GPU – let me know.

Citation:

http://brainarray.mbni.med.umich.edu/brainarray/rgpgpu/

Download

Download the gputools package for R on a Linux platform here: version 0.01.

Training on R

Here is an interesting training from Revolution Computing

New Training Course from REvolution Computing
High-Performance Computing with R
July 31, 2009 – Washington, DC – Prior to JSM
Time: 9am – 5pm
$600 commercial delegates, $450 government, $250 academic

Click Here to Register Now!

An overview of available HPC technologies for the R language to enable faster, scalable analytics that can take advantage of multiprocessor capability will be presented in a one-day course. This will include a comprehensive overview of REvolution’s recently released R packages foreach and iterators, making parallel programming easier than ever before for R programmers, as well as other available technologies such as RMPI, SNOW and many more. We will demonstrate each technology with simple examples that can be used as starting points for more sophisticated work. The agenda will also cover:

Identifying performance problems

Profiling R programs

Multithreading, using compiled code, GPGPU

Multiprocess computing

SNOW, MPI, NetWorkSpaces, and more

Batch queueing systems

Dealing with lots of data

Attendees should have basic familiarity with the R language—we will keep examples elementary but relevant to real-world applications.

This course will be conducted hands-on, classroom style. Computers will not be provided. Registrants are required to bring their own laptops.

For the full agenda Click Here or Click Here to Register Now!”

Source; www.revolution-computing.com

Disclaimer- I am NOT commerically related to REvolution, just love R. I do hope REvolution chaps do spend tiny bit of time improving the user GUI as well not just for HPC purposes.

They recently released some new packages free to the CRAN community as well

The release of 3 new packages for R designed to allow all R users to more quickly handle large, complex sets of data: iterators, foreach and doMC.

* iterators implements the “iterator” data structure familiar to users of languages like Java,

C# and Python to make it easy to program useful sequences – from all the prime numbers to the columns of a matrix or the rows of an external database.

* foreach builds on the “iterators” package to introduce a new way of programming loops in R. Unlike the traditional “for” loop, foreach runs multiple iterations simultaneously, in parallel. This makes loops run faster on a multi-core laptop, and enables distribution of large parallel-processing problems to multiple workstations in a cluster or in the cloud, without additional complicated programming. foreach works with parallel programming backends for R from the open-source and commercial domains.

* doMC is an open source parallel programming backend to enable parallel computation with “foreach” on Unix/Linux machines. It automatically enables foreach and iterator functions to work with the “multicore” package from R Core member Simon Urbanek

The new packages have been developed by REvolution Computing and released under open source licenses to the R community, enabling all existing R users

citation:

http://www.revolution-computing.com/aboutus/news-room/2009/breakthrough-parallel-packages-and-functions-for-r.php

R and SAS in Twitter Land

A tale of two languages ( set in Twitterland)

Everytime I post to the R help list, if the email contains the three words S.- A – S , I get plenty of e-spanking from senior professors and distinguished Linux people. On the other hand when I mentioned W-P-S I got dunked by the Don of SAS Global himself. We geeks are so passionate.

Here is some new stuff on Twitter for the R /Open Source community.

1) I manually made a list of

best R blogs,
R help lists ( on Nabble since Google Groups banned R help archive),
Twitter Search for #rstats ( general search word for R)

I then copied the RSS feeds of each of the above.

2) I then went to www.twitterfeed.com (uses open Id) and linked a new Twitter Account to these RSS feeds

3) I then tweaked the layout and added #rstats before each post to the new R resource http://twitter.com/Rarchive

If you are a tweeter you can follow it here http://twitter.com/Rarchive and never miss any R news going forward.

ps- I also did the same for sas for http://twitter.com/sascommunity

UPDATE

#rstats helps in SEO in Google since Google uses Twitter search as well. Existing best R search engine is http://rseek.com
In any case it is too late to change now since this is more like a automated firehose. Now you can use #rstats as well additional keywords to get more searchable useful stuff.

NOTE______

http://twitter.com/sas belongs to a guy who is wondering who is trying to hack his twitter account

, well you can check the screen shot below

Screenshot-Sky Sutton (SAS) on Twitter - Mozilla Firefox

PAW Blog Partner and 15 % off for you

paw09_blog_125

Dear Readers,

If you plan to attend Predictive Analytics World ( Oct20-21) in Washington DC,

Here are the speakers – source

Speakers Washington DC 2009:

Stephen L. Baker, Senior writer, BusinessWeek

Stephen L. Baker, author of The Numerati, is a senior writer at BusinessWeek, covering technology. Previously he was a Paris correspondent. Baker joined BusinessWeek in March, 1987, as manager of the Mexico City bureau, where he was responsible for covering Mexico and Latin America. He was named Pittsburgh bureau manager in 1992. Before BusinessWeek, Baker was a reporter for the El Paso Herald-Post. Prior to that, he was chief economic reporter for The Daily Journal in Caracas, Venezuela. Baker holds a bachelor’s degree from the University of Wisconsin and a master’s from the Columbia University Graduate School of Journalism. He blogs at TheNumerati.net and Blogspotting.net, and can be found on Twitter at @stevebaker.

John F. Elder, Ph.D., CEO and Founder, Elder Research, Inc.

Dr. John F. Elder heads a data mining consulting team with offices in Charlottesville, Virginia and Washington DC. Founded in 1995, Elder Research, Inc. focuses on scientific and commercial applications of pattern discovery and optimization, including stock selection, image recognition, text mining, biometrics, drug efficacy, credit scoring, cross-selling, investment timing, and fraud detection.

John obtained a BS and MEE in Electrical Engineering from Rice University, and a PhD in Systems Engineering from the University of Virginia, where he’s an adjunct professor, teaching Optimization or Data Mining. Prior to 13 years leading ERI, he spent 5 years in aerospace defense consulting, 4 heading research at an investment management firm, and 2 in Rice’s Computational & Applied Mathematics department.

Dr. Elder has authored innovative data mining tools, is active on Statistics, Engineering, and Finance conferences and boards, is a frequent keynote conference speaker, and is General Chair of the 2009 Knowledge Discovery and Data Mining conference in Paris. John’s courses on data analysis techniques – taught at dozens of universities, companies, and government labs – are noted for their clarity and effectiveness. Dr. Elder was honored to serve for 5 years on a panel appointed by the President to guide technology for National Security. His book on Practical Data Mining, with Bob Nisbet and Gary Minor, will appear in May 2009.

Usama Fayyad, Ph.D., CEO, Open Insights

Dr. Usama Fayyad was until recently Yahoo!’s Chief Data Officer and Executive Vice President of Research & Strategic Data Solutions where he was responsible for Yahoo!’s global data strategy, architecting Yahoo!’s data policies and systems, prioritizing data investments, and managing the Company’s data analytics and data processing infrastructure. Fayyad also founded and oversaw the Yahoo! Research organization with offices around the world. Yahoo! Research is building the premier scientific research organization to develop the new sciences of the Internet, on-line marketing, and innovative interactive applications.

Prior to joining Yahoo!, Fayyad co-founded and led the DMX Group, a data mining and data strategy consulting and technology company that was acquired by Yahoo! in 2004. In early 2000, he co-founded and served as CEO of Revenue Science, Inc.(digiMine, Inc.), a data analysis and data mining company that built, operated and hosted data warehouses and analytics for some of the world’s largest enterprises in online publishing, retail, manufacturing, telecommunications and financial services. The company today specializes in Behavioral Targeting and advertising networks. Fayyad’s professional experience also includes five years spent leading the data mining and exploration group at Microsoft Research and building the data mining products for Microsoft’s server division. From 1989 to 1996 Fayyad held a leadership role at NASA’s Jet Propulsion Laboratory (JPL), where his work in the analysis and exploration of scientific databases gathered from observatories, remote-sensing platforms and spacecraft garnered him the top research excellence award that Caltech awards to JPL scientists, as well as a U.S. Government medal from NASA.

Fayyad earned his Ph.D. in engineering from the University of Michigan, Ann Arbor (1991), and also holds BSE’s in both electrical and computer engineering (1984); MSE in computer science and engineering (1986); and M.Sc. in mathematics (1989). He has published over 100 technical articles in the fields of data mining and Artificial Intelligence, is a Fellow of the AAAI and a Fellow of the ACM, has edited two influential books on the data mining and launched and served as editor-in-chief of both the primary scientific journal in the field of data mining and the primary newsletter in the technical community published by the ACM: SIGKDD Explorations.

Eric Siegel, Ph.D., Conference Chair

The president of Prediction Impact, Inc., Eric Siegel is an expert in predictive analytics and data mining and a former computer science professor at Columbia University, where he won awards for teaching, including graduate-level courses in machine learning and intelligent systems – the academic terms for predictive analytics. After Columbia, Dr. Siegel co-founded two software companies for customer profiling and data mining, and then started Prediction Impact in 2003, providing predictive analytics services and training to mid-tier through Fortune 100 companies.

Dr. Siegel is the instructor of the acclaimed training program, Predictive Analytics for Business, Marketing and Web, and the online version, Predictive Analytics Applied. He has published 13 papers in data mining research and computer science education, has served on 10 conference program committees, and has chaired a AAAI Symposium held at MIT.

you can register at http://www.predictiveanalyticsworld.com/register.php

Here is the pricing

Pricing
Predictive Analytics World Fall 2009

Includes breakfasts, lunches, priceless networking during coffee breaks, the PAW Reception, and full access to program sessions and sponsor expositions.

	Super Early Bird Price (till June 30)	Early Bird Price (July 1 – Sept 4)	Regular Price
Two Day Pass (Oct 20-21)	$1190	$1390	$1590
Predictive Modeling Methods Workshop (Oct 22)	$695	$795	$895
Putting Predictive Analytics to Work (Oct 19)	$695	$795	$895

The discount code I can distribute to you readers is the following: BLOGDC09 (15% off a two-day pass).You can do the maths…

(Ajay- Nopes I dont get money at all in these activities as blasted by some people
- but I do hope to get some good karma. Have a good time and book now).

PAW is back

The Predictive Analytics world is going to be back in October soon , and all those who missed out the stelar event can start booking now.

Here is the official BR ( blog Release)

Source: http://www.predictiveanalyticsworld.com/blog/wp-trackback.php?p=20

June 5th 2009 10:46 am

Keynotes at October’s PAW: Stephen Baker and Usama Fayyad

Predictive Analytics World, coming October 20-21 to Washington DC, has a great line-up of keynote speakers:

Stephen Baker, author of The Numerati and senior writer at BusinessWeek, where he’s been since 1987. Steve’s book has received a tremendous amount of attention this year. It is a revealing and insightful exploration of the opportunities and pitfalls of applied analytics, and consumer perception thereof.

Usama Fayyad, Ph.D. — CEO, Open Insights and formerly Yahoo!’s Chief Data Officer and Executive Vice President of Research & Strategic Data Solutions. Dr. Fayyad will return as an acclaimed keynote speaker. His keynote at February’s PAW (San Francisco) received extremely strong ratings from conference attendees.

Finally, Eric Siegel, Ph.D., will be kicking off PAW with a reprise of his keynote, “Five Ways to Lower Costs with Predictive Analytics.”

PMML 4.0

There are some nice changes in the PMML 4.0 version. PMML is the XML version for data modeling , or specificallyquoting the DMG group itself

PMML uses XML to represent mining models. The structure of the models is described by an XML Schema. One or more mining models can be contained in a PMML document. A PMML document is an XML document with a root element of type PMML. The general structure of a PMML document is:
  <?xml version="1.0"?>
  <PMML version="4.0"
    xmlns="http://www.dmg.org/PMML-4_0"
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" >

    <Header copyright="Example.com"/>
    <DataDictionary> ... </DataDictionary>

    ... a model ...

  </PMML>

So what is new in version 4. Here are some powerful modeling changes. For anyone with any XML knowledge PMML is the way to go.

PMML 4.0 – Changes from PMML 3.2

Associations

Itemset and AssociationRule elements are no longer enclosed within a “Choice” element

Added different scoring procedures: recommendation, exclusiveRecommendation and ruleAssociation with explanation and example

Changed version to “4.0” from “3.2” in the example(s)

BuiltinFunctions

Added the following functions:

isMissing

isNotMissing

equal

notEqual

lessThan

lessOrEqual

greaterThan

greaterOrEqual

isIn

isNotIn

and

or

not

isIn

isNotIn

if

Click on Image for better resolution

ClusteringModel

Changed version to “4.0” from “3.2” in the example(s)

Added reference to ModelExplanation element in the model XSD

Conformance

Changed all version references from “3.2” to “4.0”

DataDictionary

No changes

Functions

No changes

GeneralRegression

Changed to allow for Cox survival models and model ensembles

Add new model type: CoxRegression.

Allow empty regression model when model type is CoxRegression, so that baseline-only model could be represented.

Add new optional model attributes: endTimeVariable, startTimeVariable, subjectIDVariable, statusVariable, baselineStrataVariable, modelDF.

Add optional Matrix in Predictor to specify a contrast matrix, optional attribute referencePoint in Parameter.

Add new elements: BaseCumHazardTables, EventValues, BaselineStratum, BaselineCell.

Add examples of scoring for Cox Regression and contrast matrices.

Add new type of distribution: tweedie.

Add new attribute in model: targetReferenceCategory, so that the model can be used in MiningModel.

Changed version to “4.0” from “3.2” in the example(s)

Added reference to ModelExplanation element in the model XSD

GeneralStructure

Updated the XML namespace to “http://www.dmg.org/PMML-4_0” from “http://www.dmg.org/PMML-3_2”

Added TimeSeriesModel to the PMML XSD

Header

No changes

Interoperability

Changed: “As a result, a new approach for interoperability was required and is being introduced in PMML version 3.2.” to “As a result, a new approach for interoperability was introduced in PMML version 3.2.”

MiningSchema

Added frequencyWeight and analysisWeight as new options for usageType. They will not affect scoring, but will make model information more complete.

ModelComposition — No longer used, replaced by MultipleModels

ModelExplanation

New addition to PMML 4.0 that contains information to explain the models, model fit statistics, and visualization information.

ModelVerification

No changes

MultipleModels

Replaces ModelComposition. Important additions are segmentation and ensembles.

Added reference to ModelExplanation element in the model XSD

NaïveBayes

Changed version to “4.0” from “3.2” in the example(s)

Added reference to ModelExplanation element in the model XSD

NeuralNetwork

Changed version to “4.0” from “3.2” in the example(s)

Added reference to ModelExplanation element in the model XSD

Output

Extended output type to include Association rule models. The changes add a number of new attributes: “ruleFeature”, “algorithm”, “rank”, “rankBasis”, “rankOrder” and “isMultiValued”. A new enumeration type “ruleValue” is added to the RESULT-FEATURE

Regression

Changed version to “4.0” from “3.2” in the example(s)

Added reference to ModelExplanation element in the model XSD

RuleSet

Changed version to “4.0” from “3.2” in the example(s)

Added reference to ModelExplanation element in the model XSD

Sequence

Changed version to “4.0” from “3.2” in the example(s)

Statistics

accommodate weighted counts by replacing INT-ARRAY with NUM-ARRAY in DiscrStats and ContStats

change xs:nonNegativeInteger to xs:double in several places

add new boolean attribute ‘weighted’ to UnivariateStats and PartitionFieldStats elements

add new attribute cardinality in Counts

Also some very long lines in this document are now wrapped.

SupportVectorMachine

Added optional attribute threshold

Added optional attribute classificationMethod

Attribute alternateTargetCategory removed from SupportVectorMachineModel element and moved to SupportVectorMachine element

Changed the example slightly

Changed version to “4.0” from “3.2” in the example(s)

Added reference to ModelExplanation element in the model XSD

Targets

No changes

Taxonomy

Changed: “A TableLocator may contain any description which helps an application to locate a certain table. PMML 3.2 does not yet define the content. PMML users have to use their own extensions. The same applies to InlineTable.” to “A TableLocator may contain any description which helps an application to locate a certain table. PMML standard does not yet define the content. PMML users have to use their own extensions. The same applies to InlineTable.”

Text

Changed version to “4.0” from “3.2” in the example(s)

Added reference to ModelExplanation element in the model XSD

TimeSeriesModel

New addition to PMML 4.0 to support Time series models

Transformations

No changes

TreeModel

Changed version to “4.0” from “3.2” in the example(s)

Added reference to ModelExplanation element in the model XSD

Sources

http://www.dmg.org/v4-0/GeneralStructure.html

http://www.dmg.org/v4-0/Changes.html

and here are some companies using PMML already

http://www.dmg.org/products.html

I found the tool at http://www.dmg.org/coverage/ much more interesting though (see screenshot).

Zementis who we have covered in the interviews has played a steller role in bring together this common standard for data mining. Note Kxen model is also highlighted there.

The best PMML convertor tutorial is here

http://www.zementis.com/videos/PMML_Converter_iGoogle_gadget_2_demo.htm

Please share:

Please share:

Please share:

Please share:

Speakers Washington DC 2009:

Stephen L. Baker, Senior writer, BusinessWeek

John F. Elder, Ph.D., CEO and Founder, Elder Research, Inc.

Usama Fayyad, Ph.D., CEO, Open Insights

Eric Siegel, Ph.D., Conference Chair

Pricing Predictive Analytics World Fall 2009

Two Day Pass (Oct 20-21)

Predictive Modeling Methods Workshop (Oct 22)

Putting Predictive Analytics to Work (Oct 19)

The discount code I can distribute to you readers is the following: BLOGDC09 (15% off a two-day pass).You can do the maths…

Please share:

Please share:

PMML 4.0 – Changes from PMML 3.2

Associations

BuiltinFunctions

Click on Image for better resolution

ClusteringModel

Conformance

DataDictionary

Functions

GeneralRegression

GeneralStructure

Header

Interoperability

MiningSchema

ModelComposition — No longer used, replaced by MultipleModels

ModelExplanation

ModelVerification

MultipleModels

NaïveBayes

NeuralNetwork

Output

Regression

RuleSet

Sequence

Statistics

SupportVectorMachine

Targets

Taxonomy

Text

TimeSeriesModel

Transformations

TreeModel

Please share:

Pricing
Predictive Analytics World Fall 2009

Two Day Pass
(Oct 20-21)

Predictive Modeling Methods Workshop
(Oct 22)

Putting Predictive Analytics to Work
(Oct 19)