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.

Interview Dean Abbott Abbott Analytics

Here is an interview with noted Analytics Consultant and trainer Dean Abbott. Dean is scheduled to take a workshop on Predictive Analytics at PAW (Predictive Analytics World Conference)  Oct 18 , 2010 in Washington D.C

Ajay-  Describe your upcoming hands on workshop at Predictive Analytics World and how it can help people learn more predictive modeling.

Refer- http://www.predictiveanalyticsworld.com/dc/2010/handson_predictive_analytics.php

Dean- The hands-on workshop is geared toward individuals who know something about predictive analytics but would like to experience the process. It will help people in two regards. First, by going through the data assessment, preparation, modeling and model assessment stages in one day, the attendees will see how predictive analytics works in reality, including some of the pain associated with false starts and mistakes. At the same time, they will experience success with building reasonable models to solve a problem in a single day. I have found that for many, having to actually build the predictive analytics solution if an eye-opener. Seeing demonstrations show the capabilities of a tool, but greater value for an end-user is the development of intuition of what to do at each each stage of the process that makes the theory of predictive analytics real.

Second, they will gain experience using a top-tier predictive analytics software tool, Enterprise Miner (EM). This is especially helpful for those who are considering purchasing EM, but also for those who have used open source tools and have never experienced the additional power and efficiencies that come with a tool that is well thought out from a business solutions standpoint (as opposed to an algorithm workbench).

Ajay-  You are an instructor with software ranging from SPSS, S Plus, SAS Enterprise Miner, Statistica and CART. What features of each software do you like best and are more suited for application in data cases.

Dean- I’ll add Tibco Spotfire Miner, Polyanalyst and Unica’s Predictive Insight to the list of tools I’ve taught “hands-on” courses around, and there are at least a half dozen more I demonstrate in lecture courses (JMP, Matlab, Wizwhy, R, Ggobi, RapidMiner, Orange, Weka, RandomForests and TreeNet to name a few). The development of software is a fascinating undertaking, and each tools has its own strengths and weaknesses.

I personally gravitate toward tools with data flow / icon interface because I think more that way, and I’ve tired of learning more programming languages.

Since the predictive analytics algorithms are roughly the same (backdrop is backdrop no matter which tool you use), the key differentiators are

(1) how data can be loaded in and how tightly integrated can the tool be with the database,

(2) how well big data can be handled,

(3) how extensive are the data manipulation options,

(4) how flexible are the model reporting options, and

(5) how can you get the models and/or predictions out.

There are vast differences in the tools on these matters, so when I recommend tools for customers, I usually interview them quite extensively to understand better how they use data and how the models will be integrated into their business practice.

A final consideration is related to the efficiency of using the tool: how much automation can one introduce so that user-interaction is minimized once the analytics process has been defined. While I don’t like new programming languages, scripting and programming often helps here, though some tools have a way to run the visual programming data diagram itself without converting it to code.

Ajay- What are your views on the increasing trend of consolidation and mergers and acquisitions in the predictive analytics space. Does this increase the need for vendor neutral analysts and consultants as well as conferences.

Dean- When companies buy a predictive analytics software package, it’s a mixed bag. SPSS purchasing of Clementine was ultimately good for the predictive analytics, though it took several years for SPSS to figure out what they wanted to do with it. Darwin ultimately disappeared after being purchased by Oracle, but the newer Oracle data mining tool, ODM, integrates better with the database than Darwin did or even would have been able to.

The biggest trend and pressure for the commercial vendors is the improvements in the Open Source and GNU tools. These are becoming more viable for enterprise-level customers with big data, though from what I’ve seen, they haven’t caught up with the big commercial players yet. There is great value in bringing both commercial and open source tools to the attention of end-users in the context of solutions (rather than sales) in a conference setting, which is I think an advantage that Predictive Analytics World has.

As a vendor-neutral consultant, flux is always a good thing because I have to be proficient in a variety of tools, and it is the breadth that brings value for customers entering into the predictive analytics space. But it is very difficult to keep up with the rapidly-changing market and that is something I am weighing myself: how many tools should I keep in my active toolbox.

Ajay-  Describe your career and how you came into the Predictive Analytics space. What are your views on various MS Analytics offered by Universities.

Dean- After getting a masters degree in Applied Mathematics, my first job was at a small aerospace engineering company in Charlottesville, VA called Barron Associates, Inc. (BAI); it is still in existence and doing quite well! I was working on optimal guidance algorithms for some developmental missile systems, and statistical learning was a key part of the process, so I but my teeth on pattern recognition techniques there, and frankly, that was the most interesting part of the job. In fact, most of us agreed that this was the most interesting part: John Elder (Elder Research) was the first employee at BAI, and was there at that time. Gerry Montgomery and Paul Hess were there as well and left to form a data mining company called AbTech and are still in analytics space.

After working at BAI, I had short stints at Martin Marietta Corp. and PAR Government Systems were I worked on analytics solutions in DoD, primarily radar and sonar applications. It was while at Elder Research in the 90s that began working in the commercial space more in financial and risk modeling, and then in 1999 I began working as an independent consultant.

One thing I love about this field is that the same techniques can be applied broadly, and therefore I can work on CRM, web analytics, tax and financial risk, credit scoring, survey analysis, and many more application, and cross-fertilize ideas from one domain into other domains.

Regarding MS degrees, let me first write that I am very encouraged that data mining and predictive analytics are being taught in specific class and programs rather than as just an add-on to an advanced statistics or business class. That stated, I have mixed feelings about analytics offerings at Universities.

I find that most provide a good theoretical foundation in the algorithms, but are weak in describing the entire process in a business context. For those building predictive models, the model-building stage nearly always takes much less time than getting the data ready for modeling and reporting results. These are cross-discipline tasks, requiring some understanding of the database world and the business world for us to define the target variable(s) properly and clean up the data so that the predictive analytics algorithms to work well.

The programs that have a practicum of some kind are the most useful, in my opinion. There are some certificate programs out there that have more of a business-oriented framework, and the NC State program builds an internship into the degree itself. These are positive steps in the field that I’m sure will continue as predictive analytics graduates become more in demand.

Biography-

DEAN ABBOTT is President of Abbott Analytics in San Diego, California. Mr. Abbott has over 21 years of experience applying advanced data mining, data preparation, and data visualization methods in real-world data intensive problems, including fraud detection, response modeling, survey analysis, planned giving, predictive toxicology, signal process, and missile guidance. In addition, he has developed and evaluated algorithms for use in commercial data mining and pattern recognition products, including polynomial networks, neural networks, radial basis functions, and clustering algorithms, and has consulted with data mining software companies to provide critiques and assessments of their current features and future enhancements.

Mr. Abbott is a seasoned instructor, having taught a wide range of data mining tutorials and seminars for a decade to audiences of up to 400, including DAMA, KDD, AAAI, and IEEE conferences. He is the instructor of well-regarded data mining courses, explaining concepts in language readily understood by a wide range of audiences, including analytics novices, data analysts, statisticians, and business professionals. Mr. Abbott also has taught both applied and hands-on data mining courses for major software vendors, including Clementine (SPSS, an IBM Company), Affinium Model (Unica Corporation), Statistica (StatSoft, Inc.), S-Plus and Insightful Miner (Insightful Corporation), Enterprise Miner (SAS), Tibco Spitfire Miner (Tibco), and CART (Salford Systems).