PMML Plugin for Greenplum now available

Predictive Model Markup Language
Image via Wikipedia

From a press release from Zementis.

 

, the Universal PMML Plug-in for in-database scoring. Available now for the EMC Greenplum Database, a high-performance massively parallel processing (MPP) database, the plug-in leverages the Predictive Model Markup Language (PMML) to execute predictive models directly within EMC Greenplum, for highly optimized in-database scoring.

Universal PMML Plug-in

Developed by the Data Mining Group (DMG), PMML is supported by all major data mining vendors, e.g., IBM SPSS, SAS, Teradata, FICO, STASTICA, Microstrategy, TIBCO and Revolution Analytics as well as open source tools like R, KNIME and RapidMiner. With PMML, models built in any of these data mining tools can now instantly be deployed in the EMC Greenplum database. The net result is the ability to leverage the power of standards-based predictive analytics on a massive scale, right where the data resides.

“By partnering with Zementis, a true PMML innovator, we are able to offer a vendor-agnostic solution for moving enterprise-level predictive analytics into the database execution environment,” said Dr. Steven Hillion, Vice President of Analytics at EMC Greenplum. “With Zementis and PMML, the de-facto standard for representing data mining models, we are eliminating the need to recode predictive analytic models in order to deploy them within our database. In turn, this enables an analyst to reduce the time to insight required in most businesses today.”

Want to learn more?
 

To learn more about how the EMC Greenplum Database and the Universal PMML Plug-in work together, feel free to:

  1. Visit the PMML Plug-in product page
  2. Download the white paper

The Universal PMML Plug-in for the EMC Greenplum Database is available now. Contact us today for more information.

Michael Zeller, CEO, Zementis

 

 

An Introduction to Data Mining-online book

I was reading David Smith’s blog http://blog.revolutionanalytics.com/

where he mentioned this interview of Norman Nie, at TDWI

http://tdwi.org/Articles/2010/11/17/R-101.aspx?Page=2

where I saw this link (its great if you want to study Data Mining btw)

http://www.kdnuggets.com/education/usa-canada.html

and I c/liked the U Toronto link

http://chem-eng.utoronto.ca/~datamining/

Best of All- I really liked this online book created by Professor S. Sayad

Its succinct and beautiful and describes all of the Data Mining you want to read in one Map (actually 4 images painstakingly assembled with perfection)

The best thing is- in the original map- even the sub items are click-able for specifics like Pie Chart and Stacked Column chart are not in one simple drop down like Charts- but rather by nature of the kind of variables that lead to these charts. For doing that- you would need to go to the site itself- ( see http://chem-eng.utoronto.ca/~datamining/dmc/categorical_variables.htm

vs

http://chem-eng.utoronto.ca/~datamining/dmc/categorical_numerical.htm

Again- there is no mention of the data visualization software used to create the images but I think I can take a hint from the Software Page which says software used are-

Software

See it on your own-online book (c)Professor S. Sayad

Really good DIY tutorial

http://chem-eng.utoronto.ca/~datamining/dmc/data_mining_map.htm

PAWCON -This week in London

Watch out for the twitter hash news on PAWCON and the exciting agenda lined up. If your in the City- you may want to just drop in

http://www.predictiveanalyticsworld.com/london/2010/agenda.php#day1-7

Disclaimer- PAWCON has been a blog partner with Decisionstats (since the first PAWCON ). It is vendor neutral and features open source as well proprietary software, as well case studies from academia and Industry for a balanced view.

 

Little birdie told me some exciting product enhancements may be in the works including a not yet announced R plugin 😉 and the latest SAS product using embedded analytics and Dr Elder’s full day data mining workshop.

Citation-

http://www.predictiveanalyticsworld.com/london/2010/agenda.php#day1-7

Monday November 15, 2010
All conference sessions take place in Edward 5-7

8:00am-9:00am

Registration, Coffee and Danish
Room: Albert Suites


9:00am-9:50am

Keynote
Five Ways Predictive Analytics Cuts Enterprise Risk

All business is an exercise in risk management. All organizations would benefit from measuring, tracking and computing risk as a core process, much like insurance companies do.

Predictive analytics does the trick, one customer at a time. This technology is a data-driven means to compute the risk each customer will defect, not respond to an expensive mailer, consume a retention discount even if she were not going to leave in the first place, not be targeted for a telephone solicitation that would have landed a sale, commit fraud, or become a “loss customer” such as a bad debtor or an insurance policy-holder with high claims.

In this keynote session, Dr. Eric Siegel will reveal:

  • Five ways predictive analytics evolves your enterprise to reduce risk
  • Hidden sources of risk across operational functions
  • What every business should learn from insurance companies
  • How advancements have reversed the very meaning of fraud
  • Why “man + machine” teams are greater than the sum of their parts for
  • enterprise decision support

 

Speaker: Eric Siegel, Ph.D., Program Chair, Predictive Analytics World

Top of this page ] [ Agenda overview ]


IBM9:50am-10:10am

Platinum Sponsor Presentation
The Analytical Revolution

The algorithms at the heart of predictive analytics have been around for years – in some cases for decades. But now, as we see predictive analytics move to the mainstream and become a competitive necessity for organisations in all industries, the most crucial challenges are to ensure that results can be delivered to where they can make a direct impact on outcomes and business performance, and that the application of analytics can be scaled to the most demanding enterprise requirements.

This session will look at the obstacles to successfully applying analysis at the enterprise level, and how today’s approaches and technologies can enable the true “industrialisation” of predictive analytics.

Speaker: Colin Shearer, WW Industry Solutions Leader, IBM UK Ltd

Top of this page ] [ Agenda overview ]


Deloitte10:10am-10:20am

Gold Sponsor Presentation
How Predictive Analytics is Driving Business Value

Organisations are increasingly relying on analytics to make key business decisions. Today, technology advances and the increasing need to realise competitive advantage in the market place are driving predictive analytics from the domain of marketers and tactical one-off exercises to the point where analytics are being embedded within core business processes.

During this session, Richard will share some of the focus areas where Deloitte is driving business transformation through predictive analytics, including Workforce, Brand Equity and Reputational Risk, Customer Insight and Network Analytics.

Speaker: Richard Fayers, Senior Manager, Deloitte Analytical Insight

Top of this page ] [ Agenda overview ]


10:20am-10:45am

Break / Exhibits
Room: Albert Suites


10:45am-11:35am
Healthcare
Case Study: Life Line Screening
Taking CRM Global Through Predictive Analytics

While Life Line is successfully executing a US CRM roadmap, they are also beginning this same evolution abroad. They are beginning in the UK where Merkle procured data and built a response model that is pulling responses over 30% higher than competitors. This presentation will give an overview of the US CRM roadmap, and then focus on the beginning of their strategy abroad, focusing on the data procurement they could not get anywhere else but through Merkle and the successful modeling and analytics for the UK.

Speaker: Ozgur Dogan, VP, Quantitative Solutions Group, Merkle Inc.

Speaker: Trish Mathe, Life Line Screening

Top of this page ] [ Agenda overview ]


11:35am-12:25pm
Open Source Analytics; Healthcare
Case Study: A large health care organization
The Rise of Open Source Analytics: Lowering Costs While Improving Patient Care

Rapidminer and R were the number 1 and 2 in this years annual KDNuggets data mining tool usage poll, followed by Knime on place 4 and Weka on place 6. So what’s going on here? Are these open source tools really that good or is their popularity strongly correlated with lower acquisition costs alone? This session answers these questions based on a real world case for a large health care organization and explains the risks & benefits of using open source technology. The final part of the session explains how these tools stack up against their traditional, proprietary counterparts.

Speaker: Jos van Dongen, Associate & Principal, DeltIQ Group

Top of this page ] [ Agenda overview ]


12:25pm-1:25pm

Lunch / Exhibits
Room: Albert Suites


1:25pm-2:15pm
Keynote
Thought Leader:
Case Study: Yahoo! and other large on-line e-businesses
Search Marketing and Predictive Analytics: SEM, SEO and On-line Marketing Case Studies

Search Engine Marketing is a $15B industry in the U.S. growing to double that number over the next 3 years. Worldwide the SEM market was over $50B in 2010. Not only is this a fast growing area of marketing, but it is one that has significant implications for brand and direct marketing and is undergoing rapid change with emerging channels such as mobile and social. What is unique about this area of marketing is a singularly heavy dependence on analytics:

 

  • Large numbers of variables and options
  • Real-time auctions/bids and a need to adjust strategies in real-time
  • Difficult optimization problems on allocating spend across a huge number of keywords
  • Fast-changing competitive terrain and heavy competition on the obvious channels
  • Complicated interactions between various channels and a large choice of search keyword expansion possibilities
  • Profitability and ROI analysis that are complex and often challenging

 

The size of the industry, its growing importance in marketing, its upcoming role in Mobile Advertising, and its uniquely heavy reliance on analytics makes it particularly interesting as an area for predictive analytics applications. In this session, not only will hear about some of the latest strategies and techniques to optimize search, you will hear case studies that illustrate the important role of analytics from industry practitioners.

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

Top of this page ] [ Agenda overview ]


SAS2:15pm-2:35pm

Platinum Sponsor Presentation
Creating a Model Factory Using in-Database Analytics

With the ever-increasing number of analytical models required to make fact-based decisions, as well as increasing audit compliance regulations, it is more important than ever that these models can be created, monitored, retuned and deployed as quickly and automatically as possible. This paper, using a case study from a major financial organisation, will show how organisations can build a model factory efficiently using the latest SAS technology that utilizes the power of in-database processing.

Speaker: John Spooner, Analytics Specialist, SAS (UK)

Top of this page ] [ Agenda overview ]


2:35pm-2:45pm

Session Break
Room: Albert Suites


2:45pm-3:35pm

Retail
Case Study: SABMiller
Predictive Analytics & Global Marketing Strategy

Over the last few years SABMiller plc, the second largest brewing company in the world operating in 70 countries, has been systematically segmenting its markets in different countries globally in order optimize their portfolio strategy & align it to their long term country specific growth strategy. This presentation talks about the overall methodology followed and the challenges that had to be overcome both from a technical as well as from a change management stand point in order to successfully implement a standard analytics approach to diverse markets and diverse business positions in a highly global setting.

The session explains how country specific growth strategies were converted to objective variables and consumption occasion segments were created that differentiated the market effectively by their growth potential. In addition to this the presentation will also provide a discussion on issues like:

  • The dilemmas of static vs. dynamic solutions and standardization vs. adaptable solutions
  • Challenges in acceptability, local capability development, overcoming implementation inertia, cost effectiveness, etc
  • The role that business partners at SAB and analytics service partners at AbsolutData together play in providing impactful and actionable solutions

 

Speaker: Anne Stephens, SABMiller plc

Speaker: Titir Pal, AbsolutData

Top of this page ] [ Agenda overview ]


3:35pm-4:25pm

Retail
Case Study: Overtoom Belgium
Increasing Marketing Relevance Through Personalized Targeting

 

Since many years, Overtoom Belgium – a leading B2B retailer and division of the French Manutan group – focuses on an extensive use of CRM. In this presentation, we demonstrate how Overtoom has integrated Predictive Analytics to optimize customer relationships. In this process, they employ analytics to develop answers to the key question: “which product should we offer to which customer via which channel”. We show how Overtoom gained a 10% revenue increase by replacing the existing segmentation scheme with accurate predictive response models. Additionally, we illustrate how Overtoom succeeds to deliver more relevant communications by offering personalized promotional content to every single customer, and how these personalized offers positively impact Overtoom’s conversion rates.

Speaker: Dr. Geert Verstraeten, Python Predictions

Top of this page ] [ Agenda overview ]


4:25pm-4:50pm

Break / Exhibits
Room: Albert Suites


4:50pm-5:40pm
Uplift Modelling:
Case Study: Lloyds TSB General Insurance & US Bank
Uplift Modelling: You Should Not Only Measure But Model Incremental Response

Most marketing analysts understand that measuring the impact of a marketing campaign requires a valid control group so that uplift (incremental response) can be reported. However, it is much less widely understood that the targeting models used almost everywhere do not attempt to optimize that incremental measure. That requires an uplift model.

This session will explain why a switch to uplift modelling is needed, illustrate what can and does go wrong when they are not used and the hugely positive impact they can have when used effectively. It will also discuss a range of approaches to building and assessing uplift models, from simple basic adjustments to existing modelling processes through to full-blown uplift modelling.

The talk will use Lloyds TSB General Insurance & US Bank as a case study and also illustrate real-world results from other companies and sectors.

 

Speaker: Nicholas Radcliffe, Founder and Director, Stochastic Solutions

Top of this page ] [ Agenda overview ]


5:40pm-6:30pm

Consumer services
Case Study: Canadian Automobile Association and other B2C examples
The Diminishing Marginal Returns of Variable Creation in Predictive Analytics Solutions

 

Variable Creation is the key to success in any predictive analytics exercise. Many different approaches are adopted during this process, yet there are diminishing marginal returns as the number of variables increase. Our organization conducted a case study on four existing clients to explore this so-called diminishing impact of variable creation on predictive analytics solutions. Existing predictive analytics solutions were built using our traditional variable creation process. Yet, presuming that we could exponentially increase the number of variables, we wanted to determine if this added significant benefit to the existing solution.

Speaker: Richard Boire, BoireFillerGroup

Top of this page ] [ Agenda overview ]


6:30pm-7:30pm

Reception / Exhibits
Room: Albert Suites


Tuesday November 16, 2010
All conference sessions take place in Edward 5-7

8:00am-9:00am

Registration, Coffee and Danish
Room: Albert Suites


9:00am-9:55am
Keynote
Multiple Case Studies: Anheuser-Busch, Disney, HP, HSBC, Pfizer, and others
The High ROI of Data Mining for Innovative Organizations

Data mining and advanced analytics can enhance your bottom line in three basic ways, by 1) streamlining a process, 2) eliminating the bad, or 3) highlighting the good. In rare situations, a fourth way – creating something new – is possible. But modern organizations are so effective at their core tasks that data mining usually results in an iterative, rather than transformative, improvement. Still, the impact can be dramatic.

Dr. Elder will share the story (problem, solution, and effect) of nine projects conducted over the last decade for some of America’s most innovative agencies and corporations:

    Streamline:

  • Cross-selling for HSBC
  • Image recognition for Anheuser-Busch
  • Biometric identification for Lumidigm (for Disney)
  • Optimal decisioning for Peregrine Systems (now part of Hewlett-Packard)
  • Quick decisions for the Social Security Administration
    Eliminate Bad:

  • Tax fraud detection for the IRS
  • Warranty Fraud detection for Hewlett-Packard
    Highlight Good:

  • Sector trading for WestWind Foundation
  • Drug efficacy discovery for Pharmacia & UpJohn (now Pfizer)

Moderator: Eric Siegel, Program Chair, Predictive Analytics World

Speaker: John Elder, Ph.D., Elder Research, Inc.

Also see Dr. Elder’s full-day workshop

 

Top of this page ] [ Agenda overview ]


9:55am-10:30am

Break / Exhibits
Room: Albert Suites


10:30am-11:20am
Telecommunications
Case Study: Leading Telecommunications Operator
Predictive Analytics and Efficient Fact-based Marketing

The presentation describes what are the major topics and issues when you introduce predictive analytics and how to build a Fact-Based marketing environment. The introduced tools and methodologies proved to be highly efficient in terms of improving the overall direct marketing activity and customer contact operations for the involved companies. Generally, the introduced approaches have great potential for organizations with large customer bases like Mobile Operators, Internet Giants, Media Companies, or Retail Chains.

Main Introduced Solutions:-Automated Serial Production of Predictive Models for Campaign Targeting-Automated Campaign Measurements and Tracking Solutions-Precise Product Added Value Evaluation.

Speaker: Tamer Keshi, Ph.D., Long-term contractor, T-Mobile

Speaker: Beata Kovacs, International Head of CRM Solutions, Deutsche Telekom

Top of this page ] [ Agenda overview ]


11:20am-11:25am

Session Changeover


11:25am-12:15pm
Thought Leader
Nine Laws of Data Mining

Data mining is the predictive core of predictive analytics, a business process that finds useful patterns in data through the use of business knowledge. The industry standard CRISP-DM methodology describes the process, but does not explain why the process takes the form that it does. I present nine “laws of data mining”, useful maxims for data miners, with explanations that reveal the reasons behind the surface properties of the data mining process. The nine laws have implications for predictive analytics applications: how and why it works so well, which ambitions could succeed, and which must fail.

 

Speaker: Tom Khabaza, khabaza.com

 

Top of this page ] [ Agenda overview ]


12:15pm-1:30pm

Lunch / Exhibits
Room: Albert Suites


1:30pm-2:25pm
Expert Panel: Kaboom! Predictive Analytics Hits the Mainstream

Predictive analytics has taken off, across industry sectors and across applications in marketing, fraud detection, credit scoring and beyond. Where exactly are we in the process of crossing the chasm toward pervasive deployment, and how can we ensure progress keeps up the pace and stays on target?

This expert panel will address:

  • How much of predictive analytics’ potential has been fully realized?
  • Where are the outstanding opportunities with greatest potential?
  • What are the greatest challenges faced by the industry in achieving wide scale adoption?
  • How are these challenges best overcome?

 

Panelist: John Elder, Ph.D., Elder Research, Inc.

Panelist: Colin Shearer, WW Industry Solutions Leader, IBM UK Ltd

Panelist: Udo Sglavo, Global Analytic Solutions Manager, SAS

Panel moderator: Eric Siegel, Ph.D., Program Chair, Predictive Analytics World


2:25pm-2:30pm

Session Changeover


2:30pm-3:20pm
Crowdsourcing Data Mining
Case Study: University of Melbourne, Chessmetrics
Prediction Competitions: Far More Than Just a Bit of Fun

Data modelling competitions allow companies and researchers to post a problem and have it scrutinised by the world’s best data scientists. There are an infinite number of techniques that can be applied to any modelling task but it is impossible to know at the outset which will be most effective. By exposing the problem to a wide audience, competitions are a cost effective way to reach the frontier of what is possible from a given dataset. The power of competitions is neatly illustrated by the results of a recent bioinformatics competition hosted by Kaggle. It required participants to pick markers in HIV’s genetic sequence that coincide with changes in the severity of infection. Within a week and a half, the best entry had already outdone the best methods in the scientific literature. This presentation will cover how competitions typically work, some case studies and the types of business modelling challenges that the Kaggle platform can address.

Speaker: Anthony Goldbloom, Kaggle Pty Ltd

Top of this page ] [ Agenda overview ]


3:20pm-3:50pm

Breaks /Exhibits
Room: Albert Suites


3:50pm-4:40pm
Human Resources; e-Commerce
Case Study: Naukri.com, Jeevansathi.com
Increasing Marketing ROI and Efficiency of Candidate-Search with Predictive Analytics

InfoEdge, India’s largest and most profitable online firm with a bouquet of internet properties has been Google’s biggest customer in India. Our team used predictive modeling to double our profits across multiple fronts. For Naukri.com, India’s number 1 job portal, predictive models target jobseekers most relevant to the recruiter. Analytical insights provided a deeper understanding of recruiter behaviour and informed a redesign of this product’s recruiter search functionality. This session will describe how we did it, and also reveal how Jeevansathi.com, India’s 2nd-largest matrimony portal, targets the acquisition of consumers in the market for marriage.

 

Speaker: Suvomoy Sarkar, Chief Analytics Officer, HT Media & Info Edge India (parent company of the two companies above)

 

Top of this page ] [ Agenda overview ]


4:40pm-5:00pm
Closing Remarks

Speaker: Eric Siegel, Ph.D., Program Chair, Predictive Analytics World

Top of this page ] [ Agenda overview ]


Wednesday November 17, 2010

Full-day Workshop
The Best and the Worst of Predictive Analytics:
Predictive Modeling Methods and Common Data Mining Mistakes

Click here for the detailed workshop description

  • Workshop starts at 9:00am
  • First AM Break from 10:00 – 10:15
  • Second AM Break from 11:15 – 11:30
  • Lunch from 12:30 – 1:15pm
  • First PM Break: 2:00 – 2:15
  • Second PM Break: 3:15 – 3:30
  • Workshop ends at 4:30pm

Speaker: John Elder, Ph.D., CEO and Founder, Elder Research, Inc.

 

The Top Statistical Softwares (GUI)

The list of top Statistical Softwares (GUI) is continued below. You can see the earlier post here

6. R Commander– While initially aimed at being a basic statistics GUI, the tremendous popularity of R Commander and the extensions in the form of plugins has helped make this one of the most widely used GUI. In short if you dont know ANY R, and still want to do basic descriptive stats and modeling this will come in handy- with an added script window for custom code for advanced users and extensions like that for DoE (design of experiments) and QCC (Quality Control) packages the e-plugins are a great way to extend this. I suspect the only thing holding it back is Dr Fox and the rest of R Core’s reluctance to fully embrace GUI as a software medium. You can read his earlier interview here-https://decisionstats.wordpress.com/2009/09/14/interview-professor-john-fox-creator-r-commander/

Technically it is possible to convert just about any package to a GUI menu in R Commander using the e-plugins.

7. SAS GUIs

Enterprise (Guide)

SAS Enterprise Guide was the higher end (and higher priced solution) to enhanced editor’s lack of menu driven commands. It works but many people I know prefer the text editor just as well.


The Enterprise Miner is a separate software and works more like Red R or SPSS Modeler does. Again EM is one of the major DM softwares out there, but the similarity in names is a bit confusing.

Even the Base SAS Enhanced Editor does have some menus for importing data, or querying etc, but it is rarely confused for being a GUI.

8. Oracle Data Miner and Knime

I like both the ODM and Knime but I find the lack of advertising or promotional support puzzling. Both these softwares can do well to combine technical excellence with some marketing. And since they are both free you can check them out yourself here

Oracle Data Mining

You can download it here-(note- the Oracle Web Site itself is a bit aging 🙂 )

http://www.oracle.com/technology/products/bi/odm/odminer.html

Knime is the open source GUI which can be found here-

http://www.knime.org/introduction/features

9. RAwkard

Another R GUI- it stands out on the comprehensive ways you can customize your code in menus rather than writing all or learning by rote the syntax.

From http://sourceforge.net/apps/mediawiki/rkward/index.php?title=Main_Page

you can see it below. I recommend this GUI over other GUIs especially if you are new to R and do more data visualization which needs custom graphics.

10. Red R and R JGR/ Deducer

Red R and RJGR/Deducer are both up and coming GUIs for R. While REd R is R version for Enterprise Miner, Deducer is coming up with a new GUI for ggplot the powerful graphics package in R.

Some GUIs excluded from this list are – Statistica, MatLab, EViews(?) because I dont really work with them, and thought it best to turn them over to someone who knows them better.

Hope this list of GUIs helps you- note most of the softwares can be learnt within a quick hour and two if you know basic software skills/data manipulation so going through the GUI list is a faster way of adding value to your resume/knowledge base as well.


Interview KNIME Fabian Dill

fabian We have covered KNIME.com ‘s open source platform earlier. On the eve of it’s new product launch, co-founder of Knime.com Fabian Dill reveals his thoughts in an exclusive interview.

From the Knime.com website

The modular data exploration platform KNIME, originally solely developed at the University of Konstanz, Germany, enables the user to visually create data flows – or pipelines, execute selected analysis steps, and later investigate the results through interactive views on data and models. KNIME already has more than 2,000 active users in diverse application areas, ranging from early drug discovery and customer relationship analysis to financial information integration.

Ajay – What prompted you personally to be part of KNIME and not join a big technology company? What does the future hold for KNIME in 2009-10?

Fabian -I was excited when I first joined the KNIME team in 2005. Back then, we were working exclusively on the open source version backed by some academic funding. Being part of the team that put together such a professional data mining environment from scratch was a great experience. Growing this into a commercial support and development arm has been a thrill as well. The team and the diverse experiences gained from helping get a new company off the ground and being involved in everything it takes to enable this to be successful made it unthinkable for me to work anywhere else.

We continue to develop the open source arm of KNIME and many new features lie ahead: text, image, and time series processing as well as better support for variables. We are constantly working on adding new nodes. KNIME 2.1 is expected in the fall and some of the ongoing development can already be found on the KNIME Labs page (http://labs.knime.org)

The commercial division is providing support and maintenance subscriptions for the freely available desktop version. At the same time we are developing products which will streamline the integration of KNIME into existing IT infrastructures:

  • the KNIME Grid Support lets you run your compute-intensive (sub-) workflows or nodes on a grid or cluster;

  • KNIME Reporting makes use of KNIME’s flexibility in order to gather the data for your report and provides simplified views (static or interactive=dashboards) on the resulting workflow and its results; and

  • the KNIME Enterprise Server facilitates company-wide installation of KNIME and supports collaboration between departments and sites by providing central workflow repositories, scheduled and remote execution, and user rights management.

Ajay -Software as a service and Cloud Computing is the next big thing in 2009. Are there any plans to put KNIME on a cloud computer and charge clients for the hour so they can build models on huge data without buying any hardware but just rent the time?

Fabian – Cloud computing is an agile and client-centric approach and therefore fits nicely into the KNIME framework, especially considering that we are already working on support for distributed computing of KNIME workflows (see above). However, we have no immediate plans for KNIME workflow processing on a per-use charge or similar. Thats an interesting idea, though. The way KNIME nodes are nicely encapsulated (and often even distributable themselves) would make this quite natural.

Ajay – What differentiates KNIME from other products such as RPro and Rapid Miner, for example? What are the principal challenges you have faced in developing it? Why do customers like and dislike it?

Fabian- Every tool has its strengths and weaknesses depending on the task you actually want to accomplish. The focus of KNIME is to support the user during his or her quest of understanding large and heterogeneous data and to make sense out of it. For this task, you cannot rely only on classical data mining techniques, wrapping them into a command line or otherwise configurable environment, but simple, intuitive access to those tools is required in addition to supporting visual exploration with interactive linking and brushing techniques.

By design, KNIME is a modular integration platform, which makes it easy to write own nodes (with the easy-to-use API) or integrate existing libraries or tools.

We integrated Weka, for example, because of its vast library of state-of-the-art machine learning algorithms, the open source program R – in order to provide access to a rich library of statistical functions (and of course many more) – and parts of the Chemistry Development Kit (CDK). All these integrations follow the KNIME requirements for easy and intuitive usage so the user does not need to understand the details of each tool in great depth.

A number of our commercial partners such as Schroedinger, Infocom, Symyx, Tripos, among others, also follow this paradigm and similarly integrate their tools into KNIME. Academic collaborations with ETH Zurich, Switzerland on the High Content Screening Platform HC/DC represent another positive outcome of this open architecture. We believe that this strictly result-oriented approach based on a carefully designed and professionally coded framework is a key factor of KNIME’s broad acceptance. I guess this is another big differentiator: right from the start, KNIME has been developed by a team consisting of SW developers with decades of industrial SW engineering experience.

Ajay – Any there any Asian plans for KNIME? Any other open source partnerships in the pipeline?

Fabian – We have a Japan-based partner, Infocom, who operates in the fields of life science. But we are always open for other partnerships, supporters, or collaborations.

In addition to the open source integrations mentioned above (Weka, R, CDK, HC/DC), there are many other different projects in the works and partnerships under negotiation. Keep an eye on our blog and on our Labs@KNIME page (labs.knime.org).

ABOUT

KNIME – development started in January 2004. Since then: 10 releases; approx. 350,000 lines of code; 25,000 downloads; an estimated 2000 active users. KNIME.com was founded in June 2008 in Zurich, Switzerland.

Fabian Dill – has been working for and with KNIME since 2005; co-founder of KNIME.com.

KNIME 2.0.2 released

From the makers of KNIME http://www.knime.org/blog/knime-202-released

KNIME 2.0.2 has been released and is available for download. This release includes a number of important bug fixes, amongst others addressing some chemical related issues, as well as a few new features. For further details see the detailed changelog.

In addition, a Windows 64bit version has been just released. Both versions, KNIME Desktop (win64) and KNIME SDK (win64), are still in experimental state and are intended to evaluate KNIME under Windows 64bit.

64 bit OS version is BIG milestone, though the launch did have a temporary issue with the R Plug-in being disabled

In KNIME 2.0.2, we have encountered a problem within the R Snippet (Local) node when special characters are used within the R script. Therefore we have temporarily disable the R 2.0.2 plugin from our update site. For a detailed work-around see the FAQ section.

Note KNIME has been covered before here.

KNIME and Zementis shake hands

Two very good and very customer centric (and open source ) companies shook hands on a strategic partnership today.

Knime  www.knime.org and Zementis www.zementis.com .

Decision Stats has been covering these companies and both the products are amazing good, synch in very well thanks to the support of the PMML standard and lower costs considerably for the consumer. (http://www.decisionstats.com/2009/02/knime/ ) and http://www.decisionstats.com/2009/02/interview-michael-zeller-ceozementis/ )

While Knime has both a free personal as well as a commercial license , it supports R thanks to the PMML (www.dmg.org initiative ). Knime also supports R very well .

See http://www.knime.org/blog/export-and-convert-r-models-pmml-within-knime

The following example R script learns a decision tree based on the Iris-Data and exports this as PMML and as an R model which is understood by the R Predictor node:

# load the library for learning a tree model
library(rpart);
# load the pmml export library
library(pmml);
# use class column as predicted column to build decision tree
dt <- rpart(class~., R)
# export to PMML
r_pmml <- pmml(dt)
# write the PMML model to an export file
write(toString(r_pmml), file="C:/R.pmml")
# provide the native R model at the out-port
R<-dt

 

Zementis takes the total cost of ownership and total pain of creating scored models to something close to 1$ /hour thanks to using their proprietary ADAPA engine.