Tag: modeling

New Plotters in Rapid Miner 5.2

I almost missed this because of my vacation and traveling

Rapid Miner has a tonne of new stuff (Statuary Ethics Declaration- Rapid Miner has been an advertising partner for Decisionstats – see the right margin)

see

http://rapid-i.com/component/option,com_myblog/Itemid,172/lang,en/

Great New Graphical Plotters

and some flashy work

and a great series of educational lectures

A Simple Explanation of Decision Tree Modeling based on Entropies

Link: http://www.simafore.com/blog/bid/94454/A-simple-explanation-of-how-entropy-fuels-a-decision-tree-model

Description of some of the basics of decision trees. Simple and hardly any math, I like the plots explaining the basic idea of the entropy as splitting criterion (although we actually calculate gain ratio differently than explained…)

Logistic Regression for Business Analytics using RapidMiner

Link: http://www.simafore.com/blog/bid/57924/Logistic-regression-for-business-analytics-using-RapidMiner-Part-2

Same as above, but this time for modeling with logistic regression.
Easy to read and covering all basic ideas together with some examples. If you are not familiar with the topic yet, part 1 (see below) might help.

Part 1 (Basics): http://www.simafore.com/blog/bid/57801/Logistic-regression-for-business-analytics-using-RapidMiner-Part-1

Deploy Model: http://www.simafore.com/blog/bid/82024/How-to-deploy-a-logistic-regression-model-using-RapidMiner

Advanced Information: http://www.simafore.com/blog/bid/99443/Understand-3-critical-steps-in-developing-logistic-regression-models

and lastly a new research project for collaborative data mining

http://www.e-lico.eu/

e-LICO Architecture and Components

The goal of the e-LICO project is to build a virtual laboratory for interdisciplinary collaborative research in data mining and data-intensive sciences. The proposed e-lab will comprise three layers: the e-science and data mining layers will form a generic research environment that can be adapted to different scientific domains by customizing the application layer.

  1. Drag a data set into one of the slots. It will be automatically detected as training data, test data or apply data, depending on whether it has a label or not.
  2. Select a goal. The most frequent one is probably “Predictive Modelling”. All goals have comments, so you see what they can be used for.
  3. Select “Fetch plans” and wait a bit to get a list of processes that solve your problem. Once the planning completes, select one of the processes (you can see a preview at the right) and run it. Alternatively, select multiple (selecting none means selecting all) and evaluate them on your data in a batch.

The assistant strives to generate processes that are compatible with your data. To do so, it performs a lot of clever operations, e.g., it automatically replaces missing values if missing values exist and this is required by the learning algorithm or performs a normalization when using a distance-based learner.

You can install the extension directly by using the Rapid-I Marketplace instead of the old update server. Just go to the preferences and enter http://rapidupdate.de:8180/UpdateServer as the update URL

Of course Rapid Miner has been of the most professional open source analytics company and they have been doing it for a long time now. I am particularly impressed by the product map (see below) and the graphical user interface.

http://rapid-i.com/content/view/186/191/lang,en/

Product Map

Just click on the products in the overview below in order to get more information about Rapid-I products.

 

Rapid-I Product Overview 

 

Quantitative Modeling for Arbitrage Positions in Ad KeyWords Internet Marketing

Assume you treat an ad keyword as an equity stock. There are slight differences in the cost for advertising for that keyword across various locations (Zurich vs Delhi) and various channels (Facebook vs Google) . You get revenue if your website ranks naturally in organic search for the keyword, and you have to pay costs for getting traffic to your website for that keyword.
An arbitrage position is defined as a riskless profit when cost of keyword is less than revenue from keyword. We take examples of Adsense  and Adwords primarily.
There are primarily two types of economic curves on the foundation of which commerce of the  internet  resides-
1) Cost Curve- Cost of Advertising to drive traffic into the website  (Google Adwords, Twitter Ads, Facebook , LinkedIn ads)
2) Revenue Curve – Revenue from ads clicked by the incoming traffic on website (like Adsense, LinkAds, Banner Ads, Ad Sharing Programs , In Game Ads)
The cost and revenue curves are primarily dependent on two things
1) Type of KeyWord-Also subdependent on
a) Location of Prospective Customer, and
b) Net Present Value of Good and Service to be eventually purchased
For example , keyword for targeting sales of enterprise “business intelligence software” should ideally be costing say X times as much as keywords for “flower shop for birthdays” where X is the multiple of the expected payoffs from sales of business intelligence software divided by expected payoff from sales of flowers (say in Location, Daytona Beach ,Florida or Austin, Texas)
2) Traffic Volume – Also sub-dependent on Time Series and
a) Seasonality -Annual Shoppping Cycle
b) Cyclicality– Macro economic shifts in time series
The cost and revenue curves are not linear and ideally should be continuous in a definitive exponential or polynomial manner, but in actual reality they may have sharp inflections , due to location, time, as well as web traffic volume thresholds
Type of Keyword – For example ,keywords for targeting sales for Eminem Albums may shoot up in a non linear manner after the musician dies.
The third and not so publicly known component of both the cost and revenue curves is factoring in internet industry dynamics , including relative market share of internet advertising platforms, as well as percentage splits between content creator and ad providing platforms.
For example, based on internet advertising spend, people belive that the internet advertising is currently heading for a duo-poly with Google and Facebook are the top two players, while Microsoft/Skype/Yahoo and LinkedIn/Twitter offer niche options, but primarily depend on price setting from Google/Bing/Facebook.
It is difficut to quantify  the elasticity and efficiency of market curves as most literature and research on this is by in-house corporate teams , or advisors or mentors or consultants to the primary leaders in a kind of incesteous fraternal hold on public academic research on this.
It is recommended that-
1) a balance be found in the need for corporate secrecy to protest shareholder value /stakeholder value maximization versus the need for data liberation for innovation and grow the internet ad pie faster-
2) Cost and Revenue Curves between different keywords, time,location, service providers, be studied by quants for hedging inetrent ad inventory or /and choose arbitrage positions This kind of analysis is done for groups of stocks and commodities in the financial world, but as commerce grows on the internet this may need more specific and independent quants.
3) attention be made to how cost and revenue curves mature as per level of sophistication of underlying economy like Brazil, Russia, China, Korea, US, Sweden may be in different stages of internet ad market evolution.
For example-
A study in cost and revenue curves for certain keywords across domains across various ad providers across various locations from 2003-2008 can help academia and research (much more than top ten lists of popular terms like non quantitative reports) as well as ensure that current algorithmic wightings are not inadvertently given away.
Part 2- of this series will explore the ways to create third party re-sellers of keywords and measuring impacts of search and ad engine optimization based on keywords.

Topic Models

Some stuff on Topic Models-

http://en.wikipedia.org/wiki/Topic_model

In machine learning and natural language processing, a topic model is a type of statistical model for discovering the abstract “topics” that occur in a collection of documents. An early topic model was probabilistic latent semantic indexing (PLSI), created by Thomas Hofmann in 1999.[1] Latent Dirichlet allocation (LDA), perhaps the most common topic model currently in use, is a generalization of PLSI developed by David Blei, Andrew Ng, and Michael Jordan in 2002, allowing documents to have a mixture of topics.[2] Other topic models are generally extensions on LDA, such as Pachinko allocation, which improves on LDA by modeling correlations between topics in addition to the word correlations which constitute topics. Although topic models were first described and implemented in the context of natural language processing, they have applications in other fields such as bioinformatics.

http://en.wikipedia.org/wiki/Latent_Dirichlet_allocation

In statistics, latent Dirichlet allocation (LDA) is a generative model that allows sets of observations to be explained by unobserved groups that explain why some parts of the data are similar. For example, if observations are words collected into documents, it posits that each document is a mixture of a small number of topics and that each word’s creation is attributable to one of the document’s topics. LDA is an example of a topic model

David M Blei’s page on Topic Models-

http://www.cs.princeton.edu/~blei/topicmodeling.html

The topic models mailing list is a good forum for discussing topic modeling.

In R,

Some resources I compiled on Slideshare based on the above- Continue reading “Topic Models”

Interview Dan Steinberg Founder Salford Systems

Here is an interview with Dan Steinberg, Founder and President of Salford Systems (http://www.salford-systems.com/ )

Ajay- Describe your journey from academia to technology entrepreneurship. What are the key milestones or turning points that you remember.

 Dan- When I was in graduate school studying econometrics at Harvard,  a number of distinguished professors at Harvard (and MIT) were actively involved in substantial real world activities.  Professors that I interacted with, or studied with, or whose software I used became involved in the creation of such companies as Sun Microsystems, Data Resources, Inc. or were heavily involved in business consulting through their own companies or other influential consultants.  Some not involved in private sector consulting took on substantial roles in government such as membership on the President’s Council of Economic Advisors. The atmosphere was one that encouraged free movement between academia and the private sector so the idea of forming a consulting and software company was quite natural and did not seem in any way inconsistent with being devoted to the advancement of science.

 Ajay- What are the latest products by Salford Systems? Any future product plans or modification to work on Big Data analytics, mobile computing and cloud computing.

 Dan- Our central set of data mining technologies are CART, MARS, TreeNet, RandomForests, and PRIM, and we have always maintained feature rich logistic regression and linear regression modules. In our latest release scheduled for January 2012 we will be including a new data mining approach to linear and logistic regression allowing for the rapid processing of massive numbers of predictors (e.g., one million columns), with powerful predictor selection and coefficient shrinkage. The new methods allow not only classic techniques such as ridge and lasso regression, but also sub-lasso model sizes. Clear tradeoff diagrams between model complexity (number of predictors) and predictive accuracy allow the modeler to select an ideal balance suitable for their requirements.

The new version of our data mining suite, Salford Predictive Modeler (SPM), also includes two important extensions to the boosted tree technology at the heart of TreeNet.  The first, Importance Sampled learning Ensembles (ISLE), is used for the compression of TreeNet tree ensembles. Starting with, say, a 1,000 tree ensemble, the ISLE compression might well reduce this down to 200 reweighted trees. Such compression will be valuable when models need to be executed in real time. The compression rate is always under the modeler’s control, meaning that if a deployed model may only contain, say, 30 trees, then the compression will deliver an optimal 30-tree weighted ensemble. Needless to say, compression of tree ensembles should be expected to be lossy and how much accuracy is lost when extreme compression is desired will vary from case to case. Prior to ISLE, practitioners have simply truncated the ensemble to the maximum allowable size.  The new methodology will substantially outperform truncation.

The second major advance is RULEFIT, a rule extraction engine that starts with a TreeNet model and decomposes it into the most interesting and predictive rules. RULEFIT is also a tree ensemble post-processor and offers the possibility of improving on the original TreeNet predictive performance. One can think of the rule extraction as an alternative way to explain and interpret an otherwise complex multi-tree model. The rules extracted are similar conceptually to the terminal nodes of a CART tree but the various rules will not refer to mutually exclusive regions of the data.

 Ajay- You have led teams that have won multiple data mining competitions. What are some of your favorite techniques or approaches to a data mining problem.

 Dan- We only enter competitions involving problems for which our technology is suitable, generally, classification and regression. In these areas, we are  partial to TreeNet because it is such a capable and robust learning machine. However, we always find great value in analyzing many aspects of a data set with CART, especially when we require a compact and easy to understand story about the data. CART is exceptionally well suited to the discovery of errors in data, often revealing errors created by the competition organizers themselves. More than once, our reports of data problems have been responsible for the competition organizer’s decision to issue a corrected version of the data and we have been the only group to discover the problem.

In general, tackling a data mining competition is no different than tackling any analytical challenge. You must start with a solid conceptual grasp of the problem and the actual objectives, and the nature and limitations of the data. Following that comes feature extraction, the selection of a modeling strategy (or strategies), and then extensive experimentation to learn what works best.

 Ajay- I know you have created your own software. But are there other software that you use or liked to use?

 Dan- For analytics we frequently test open source software to make sure that our tools will in fact deliver the superior performance we advertise. In general, if a problem clearly requires technology other than that offered by Salford, we advise clients to seek other consultants expert in that other technology.

 Ajay- Your software is installed at 3500 sites including 400 universities as per http://www.salford-systems.com/company/aboutus/index.html What is the key to managing and keeping so many customers happy?

 Dan- First, we have taken great pains to make our software reliable and we make every effort  to avoid problems related to bugs.  Our testing procedures are extensive and we have experts dedicated to stress-testing software . Second, our interface is designed to be natural, intuitive, and easy to use, so the challenges to the new user are minimized. Also, clear documentation, help files, and training videos round out how we allow the user to look after themselves. Should a client need to contact us we try to achieve 24-hour turn around on tech support issues and monitor all tech support activity to ensure timeliness, accuracy, and helpfulness of our responses. WebEx/GotoMeeting and other internet based contact permit real time interaction.

 Ajay- What do you do to relax and unwind?

 Dan- I am in the gym almost every day combining weight and cardio training. No matter how tired I am before the workout I always come out energized so locating a good gym during my extensive travels is a must. I am also actively learning Portuguese so I look to watch a Brazilian TV show or Portuguese dubbed movie when I have time; I almost never watch any form of video unless it is available in Portuguese.

 Biography-

http://www.salford-systems.com/blog/dan-steinberg.html

Dan Steinberg, President and Founder of Salford Systems, is a well-respected member of the statistics and econometrics communities. In 1992, he developed the first PC-based implementation of the original CART procedure, working in concert with Leo Breiman, Richard Olshen, Charles Stone and Jerome Friedman. In addition, he has provided consulting services on a number of biomedical and market research projects, which have sparked further innovations in the CART program and methodology.

Dr. Steinberg received his Ph.D. in Economics from Harvard University, and has given full day presentations on data mining for the American Marketing Association, the Direct Marketing Association and the American Statistical Association. After earning a PhD in Econometrics at Harvard Steinberg began his professional career as a Member of the Technical Staff at Bell Labs, Murray Hill, and then as Assistant Professor of Economics at the University of California, San Diego. A book he co-authored on Classification and Regression Trees was awarded the 1999 Nikkei Quality Control Literature Prize in Japan for excellence in statistical literature promoting the improvement of industrial quality control and management.

His consulting experience at Salford Systems has included complex modeling projects for major banks worldwide, including Citibank, Chase, American Express, Credit Suisse, and has included projects in Europe, Australia, New Zealand, Malaysia, Korea, Japan and Brazil. Steinberg led the teams that won first place awards in the KDDCup 2000, and the 2002 Duke/TeraData Churn modeling competition, and the teams that won awards in the PAKDD competitions of 2006 and 2007. He has published papers in economics, econometrics, computer science journals, and contributes actively to the ongoing research and development at Salford.

Analytics 2011 Conference

From http://www.sas.com/events/analytics/us/

The Analytics 2011 Conference Series combines the power of SAS’s M2010 Data Mining Conference and F2010 Business Forecasting Conference into one conference covering the latest trends and techniques in the field of analytics. Analytics 2011 Conference Series brings the brightest minds in the field of analytics together with hundreds of analytics practitioners. Join us as these leading conferences change names and locations. At Analytics 2011, you’ll learn through a series of case studies, technical presentations and hands-on training. If you are in the field of analytics, this is one conference you can’t afford to miss.

Conference Details

October 24-25, 2011
Grande Lakes Resort
Orlando, FL

Analytics 2011 topic areas include:

Predictive Analytics World

Here is an announcement from Predictive Analytics World, the worlds largest vendor neutral conference dedicated to Predictive Analytics alone. Decisionstats has been a blog partner of PAWCON since inception. This is cool stuff!Predictive Analytics World New York October 2011

Video Testimonials: Reasons to Attend Predictive Analytics World Oct 2011, NY  

What’s Predictive Analytics World (PAW) all about and why should you go? See and hear experiences from those who have attended PAW. The video recorded at PAW San Francisco 2011 includes statements from Thomas Davenport, conference chair Eric Siegel, and other conference participants and VIPs.

 

Join your peers October 17-21, 2011 at the Hilton New York for Predictive Analytics World, the business event for predictive analytics professionals, managers and commercial practitioners, covering today’s commercial deployment of predictive analytics, across industries and across software vendors.

Register using the code REDC before June 15th and 10% of your registration proceeds will be donated to American Red Cross Midwest Tornado Relief Effort. Also, take advantage of Super Early Bird Pricing and realize $400 in savings.

Discover new content covering all the latest topics and advanced methods by participating in PAW’s workshops, case studies, and educational sessions.   View full agenda and topics online now.

PAW NYC agenda highlights include:

  • Keynotes from Tom Davenport, President’s Distinguished Professor, Babson College, Author, Competing on Analytics and Eric Siegel,  Conference Program Chair, Predictive Analytics World
  • Special plenary sessions from industry heavyweights, Usama Fayyad, Ph.D., CEO, Open Insights and John F. Elder, CEO and Founder, Elder Research
  • Full day workshops that cover the topics of Decisioning, Core Methods, Net Lift Modeling, Hands-On Intro, Hands-On R, Intro to Predictive Analytics and Intro to Business Analytics
  • Topics covering black box trading, churn modeling, crowdsourcing, demand forecasting, ensemble models, fraud detection, healthcare, insurance applications, law enforcement, litigation, market mix modeling, mobile analytics, online marketing, risk management, social data, supply chain management, targeting direct marketing, uplift modeling (net lift), and other innovative applications that benefit organizations in new and creative ways.
Thomas Davenport
Thomas Davenport
Author, Competing on Analytics		 Eric Siegel, Ph.D
VIP from IBM Research (TBA)
Keynote on Jeopardy-Winning Watson and DeepQA		 Eric Siegel, Ph.D
Eric Siegel, Ph.D
Program Chair, Predictive Analytics World
Usama Fayyad, Ph.D
Usama Fayyad, Ph.D
CEO, Open Insights		 John F. Elder IV, Ph.D
John F. Elder IV, Ph.D
Chief Scientist, Elder Research, Inc.

Become an invaluable resource to your organization by discovering new processes and tactics that your peers are using to optimize with the best methods that leverage data – bringing their business results to the next level.

New Financial Services Track — You Asked and We Delivered

October’s event will include a new conference track of sessions dedicated to the Financial Services industry. This track will feature something for users of all levels, whether you’re deploying your first initiative or learning new ways to position analytics within your organization.

Text analytics. The new conference Text Analytics World,
co-located with PAW NYC, complements PAW’s agenda
with reasonable cross-registration options.

Take advantage of Super Early Bird Pricing and realize
$400 in savings before June 15, 2011.

Note:  Each additional attendee from the same company registered at the same time receives an extra $200 off the Conference Pass.

Register Now!

eMetrics New York
Follow Us on Twitter Be a Fan on Facebook LinkedIn Group Live Twitter Feed
Follow Like Join Live

Register Now!
All Analytics Conferences: 

Predictive Analytics World for Government – Sept 12-13 in DC
Predictive Analytics World NYC – Oct 17-21
Text Analytics World NYC – Oct 19-20
Predictive Analytics World San Francisco – March 2012
Predictive Analytics World Videos – Available on-demand
Produced by: 

Predictionimpact
RisingMedia

 

RapidMiner launches extensions marketplace

For some time now, I had been hoping for a place where new package or algorithm developers get at least a fraction of the money that iPad or iPhone application developers get. Rapid Miner has taken the lead in establishing a marketplace for extensions. Is there going to be paid extensions as well- I hope so!!

This probably makes it the first “app” marketplace in open source and the second app marketplace in analytics after salesforce.com

It is hard work to think of new algols, and some of them can really be usefull.

Can we hope for #rstats marketplace where people downloading say ggplot3.0 atleast get a prompt to donate 99 cents per download to Hadley Wickham’s Amazon wishlist. http://www.amazon.com/gp/registry/1Y65N3VFA613B

Do you think it is okay to pay 99 cents per iTunes song, but not pay a cent for open source software.

I dont know- but I am just a capitalist born in a country that was socialist for the first 13 years of my life. Congratulations once again to Rapid Miner for innovating and leading the way.

http://rapid-i.com/component/option,com_myblog/show,Rapid-I-Marketplace-Launched.html/Itemid,172

RapidMiner, Marketplace, Extensions 30 May 2011
Rapid-I Marketplace Launched by Simon Fischer

Over the years, many of you have been developing new RapidMiner Extensions dedicated to a broad set of topics. Whereas these extensions are easy to install in RapidMiner – just download and place them in the plugins folder – the hard part is to find them in the vastness that is the Internet. Extensions made by ourselves at Rapid-I, on the other hand,  are distributed by the update server making them searchable and installable directly inside RapidMiner.

We thought that this was a bit unfair, so we decieded to open up the update server to the public, and not only this, we even gave it a new look and name. The Rapid-I Marketplace is available in beta mode at http://rapidupdate.de:8180/ . You can use the Web interface to browse, comment, and rate the extensions, and you can use the update functionality in RapidMiner by going to the preferences and entering http://rapidupdate.de:8180/UpdateServer/ as the update server URL. (Once the beta test is complete, we will change the port back to 80 so we won’t have any firewall problems.)

As an Extension developer, just register with the Marketplace and drop me an email (fischer at rapid-i dot com) so I can give you permissions to upload your own extension. Upload is simple provided you use the standard RapidMiner Extension build process and will boost visibility of your extension.

Looking forward to see many new extensions there soon!

Disclaimer- Decisionstats is a partner of Rapid Miner. I have been liking the software for a long long time, and recently agreed to partner with them just like I did with KXEN some years back, and with Predictive AnalyticsConference, and Aster Data until last year.

I still think Rapid Miner is a very very good software,and a globally created software after SAP.

Here is the actual marketplace

http://rapidupdate.de:8180/UpdateServer/faces/index.xhtml

Welcome to the Rapid-I Marketplace Public Beta Test

The Rapid-I Marketplace will soon replace the RapidMiner update server. Using this marketplace, you can share your RapidMiner extensions and make them available for download by the community of RapidMiner users. Currently, we are beta testing this server. If you want to use this server in RapidMiner, you must go to the preferences and enter http://rapidupdate.de:8180/UpdateServer for the update url. After the beta test, we will change the port back to 80, which is currently occupied by the old update server. You can test the marketplace as a user (downloading extensions) and as an Extension developer. If you want to publish your extension here, please let us know via the contact form.

Hot Downloads
«« « 1 2 3 » »»
[Icon]The Image Processing Extension provides operators for handling image data. You can extract attributes describing colour and texture in the image, you can make several transformation of a image data which allows you to perform segmentation and detection of suspicious areas in image data.The extension provides many of image transformation and extraction operators ranging from Wavelet Decomposition, Hough Circle to Block Difference of Inverse probabilities.
[Icon]RapidMiner is unquestionably the world-leading open-source system for data mining. It is available as a stand-alone application for data analysis and as a data mining engine for the integration into own products. Thousands of applications of RapidMiner in more than 40 countries give their users a competitive edge.

  • Data Integration, Analytical ETL, Data Analysis, and Reporting in one single suite
  • Powerful but intuitive graphical user interface for the design of analysis processes
  • Repositories for process, data and meta data handling
  • Only solution with meta data transformation: forget trial and error and inspect results already during design time
  • Only solution which supports on-the-fly error recognition and quick fixes
  • Complete and flexible: Hundreds of data loading, data transformation, data modeling, and data visualization methods
[Icon]All modeling methods and attribute evaluation methods from the Weka machine learning library are available within RapidMiner. After installing this extension you will get access to about 100 additional modelling schemes including additional decision trees, rule learners and regression estimators.This extension combines two of the most widely used open source data mining solutions. By installing it, you can extend RapidMiner to everything what is possible with Weka while keeping the full analysis, preprocessing, and visualization power of RapidMiner.
[Icon]Finally, the two most widely used data analysis solutions – RapidMiner and R – are connected. Arbitrary R models and scripts can now be directly integrated into the RapidMiner analysis processes. The new R perspective offers the known R console together with the great plotting facilities of R. All variables and R scripts can be organized in the RapidMiner Repository.A directly included online help and multi-line editing makes the creation of R scripts much more comfortable.
Related articles
%d bloggers like this: