This spring I was contacted by Holly Thorpe of the Wenatchee Valley Business World about doing a piece on Eigenvector. I was, of course, pleased to oblige! The timing was especially good as we celebrated our 20th Anniversary earlier this year. The resulting article captures the flavor of EVRI quite well, and is a great layperson’s introduction to the type of work we do.

Our thanks go to Holly for the article and photographer Don Seabrook for capturing my work environment. (My daughter commented that she had never seen my office that clean!) I really enjoyed the process–I always get really enthusiastic when I start talking about Eigenvector!


Each year, Tuesday evening at Eigenvector University features the PLS_Toolbox/Solo User Poster Session. Prizes are given for the two best posters as judged by the Eigenvector staff, associates, and guest instructors.

At EigenU 2014 there were 10 posters presented by users, plus a couple more from EVRI, which made for a very lively session. Associate Rasmus Bro and guest lecturer Age Smilde, (who was there to lead the “Chemometrics in Metabolomics” course), assisted the Eigenvectorians in choosing the best posters. The winners were Eric Massicotte of the Canada Border Services Agency (CBSA) and Sarah Nielsen of Janssen Supply Group.

Massicotte displayed “Determination of the Geographic Origin of Tobacco Leaves by NIR.” A Partial Least Squares Discriminant Analysis (PLS-DA) model based on the NIR spectra of flue-cured tobacco leaves was developed that classified the samples into the 10 countries of origin. The sensitivity and specificity of the model was found to be greater than 0.92 for all classes except the USA group, which has the largest variability. The technique may be used in the future as a screening tool by the CBSA.

Nielsen presented “Distribution Analysis of Components in the EVRA® Patch by Confocal Raman Microscopy” which she developed with her co-author Amber Mantz. Confocal Raman Microscopy (CRM) images were generated at various locations within EVRA® samples to charaterize drug product and excipient distributions. PLS models were developed to predict relative drug concentration in the patches. Samples were analyzed from time of manufacture to product expiry to determine if changes in drug concentrations occurred. The models indicated that there were no significant changes over the life of the product.

Nielsen and Massicotte each received an Apple iPod nanos engraved with Eigenvector University 2014 Best Poster for their efforts. Good work, and thanks to everybody that presented at EigenU 2014!


Eigenvector Research, Inc. (EVRI) has an opening for a full time Chemometrics Consultant staff member. Applicants should have a M.S. or Ph.D. degree in Chemistry, Chemical Engineering, or a closely related field and at least five years experience developing chemometric models. Experience in the pharmaceutical, chemical, medical device, food and beverage or bioinformatics fields would be useful. Practical knowledge of analytical techniques, especially spectroscopy, required. Must be proficient in MATLAB. Experience with Javascript, HTML, CSS and other programming languages would be a plus. Experience with EVRI’s software, especially PLS_Toolbox and MIA_Toolbox, sought. Experience with other chemometrics packages also a plus. Teaching experience desired.

The successful candidate is expected to take a lead role on some chemometrics consulting projects and a supporting role on others. Must be able to meet with potential clients, understand their goals and needs, develop a statement of work, and execute the tasks. Good written and oral communications skills, especially the ability to convey complex information to non-experts, required. Potential to attract new projects through new and existing contacts preferred. Must be able to work at home. Location not critical but proximity to our existing locations would be a plus.

EVRI employees enjoy working on interesting projects with a dedicated, fun and lively team of chemometrics and programming experts. EVRI offers a competitive salary and benefits package, plus flexible hours and the ability to work at home.

Applicants should send a C.V. and letter of interest to Barry M. Wise.

19th Feb, 2015

Eigenvector Turns 20

Eigenvector Research, Inc. (aka EVRI) passed a significant milestone when we turned 20 years old on January 1, 2015. The event passed without a lot of fanfare around here. Frankly we were just all a little bit too busy to do much beyond offering a few internal congratulations!

Eigenvector was founded on January 1, 1995 when Neal Gallagher and I used our credit cards to buy computers and office furniture. Neal was living with us in West Richland, WA, and that became our first offices. Our original plan was that we would be primarily a consulting company and also continue to develop our MATLAB-based PLS_Toolbox. Our first two consulting projects were for Pacific Northwest National Laboratory (PNNL) where we had both been employed. The first involved development of models based on FTIR spectroscopy to estimate the organic gas content and ozone forming potential of automobile exhaust. The second was analysis of the behavior of waste tank SY-101, Hanford’s notorious “burping” tank.

We soon discovered that there was a significant demand for chemometrics courses. The field was relatively new and lots of companies wanted to take advantage of the possibilities that multivariate analysis opened up. I had taught a chemometrics course at the Joint Center for Graduate Study, (now Washington State University Tri-Cities), and was happy to take it on the road. Thus we became the three-legged stool that we are today, providing chemometric software, consulting and training.

In 2001 we made the first addition to our technical staff with Jeremy M. Shaver, now our Chief of Technology Development. This was possibly the best decision we’ve ever made. Under Jeremy’s stewardship PLS_Toolbox has developed into the user-friendly and comprehensive package that it is today. Plus we’ve added our stand-alone Solo, tools for Multivariate Image Analysis, and on-line predictor packages Solo_Predictor and Model_Exporter.

Our software sales have grown continuously for more than a decade and now comprise the largest third of our business. We now have 2500+ users in more than 1000 companies and universities across 50+ countries.

We joke sometimes that we founded EVRI because life is too short to drink bad beer, do boring work or live in a crappy place. But there’s an element of truth in that. We’ve gotten to work on some fascinating projects over the last 20 years. Some frustrating ones as well, but on average pretty interesting stuff. We’d like to think we’ve done some good in that we’ve trained a lot of chemometricians, provided tools that make their work easier, and advanced projects that have made processes more efficient and benefited consumers and patients.

We look forward to continued growth and more stimulating projects over the next 20 years!


Model_Exporter is EVRI’s software for turning multivariate/chemometric models into formats which can be compiled into online applications. It offers an alternative to our stand-alone prediction engine Solo_Predictor. Model_Exporter allows users of our MATLAB® based PLS_Toolbox and stand-alone Solo to easily create a numerical recipes of their models. These recipes give the step by step procedure that take a measurement and calculate the desired outputs, such as concentration, class assignment, prediction diagnostics, etc. This includes applying all preprocessing steps along with the model (PCA, PLS, PLS-DA etc.) itself. When Model_Exporter is installed, models can be exported into predictor files in a variety of formats via the file menu in the Analysis window as shown below.


Model_Exporter also includes two versions of the freely-distributable Model_Interpreter. Either the C# or Java version of the Model_Interpreter can be used by any 3rd party program to add the ability to parse an exported model in XML format. Simply point the interpreter at an XML exported model and supply the data from which to make a prediction. The interpreter applies the model and returns the results. Model_Interpreter has no licensing fees and is appropriate for use on standard processors and operating systems or on handheld devices run by reduced instruction set processors (e.g. ARM). Your application doesn’t need to know anything about the preprocessing or model being used.

Version 3.0 of Model_Exporter was released in early October along with its associated stand-alone Solo+Model_Exporter version 7.9. This release includes support for Support Vector Machine (SVM) regression and classification models as well as Artificial Neural Network (ANN) regression models.

These changes represent a significant addition to Model_Exporter making it even more unique in the chemometrics world. No other chemometric modeling product offers anything as transparent, flexible or unencumbered by licensing. You can get more info about Model_Exporter by consulting the Release Notes and the Model_Exporter Wiki page.

Users with current maintenance can access these versions now from their account. If expired, maintenance can be renewed through the “Purchase” tab.

If you have any questions, feel free to write us at


In the last several years we’ve seen a resurgence of interest in Classical Least Squares (CLS) modeling. To address that our Neal Gallagher is developing a course on CLS Methods for the next EigenU. Our interest also stems from the fact that we’ve worked on a number of consulting projects where CLS models are appropriate for calibrating spectroscopic systems. As you might expect, these systems are relatively simple mixtures in gas or liquid phase. Recall the CLS model is

X = CS‘ + E

where X is the measured spectra, C is the matrix of concentrations, S is the pure component spectra and E is noise.

Complicating matters a bit, several of the systems we’ve worked with exhibit significant nonlinearities due to high absorbance features. In spite of that, CLS models can work quite well if set up correctly. What follows is an example that demonstrates this (which I originally did just to clarify how this works in my own mind).

Suppose you have a single component system with a pure component response that is a simple Gaussian peak centered in the spectral range with a maximum value of one when the concentration is also one. Furthermore, suppose that the spectra is linear up to an absorbance of one but rolls off after that. (For xideal > 1 I used xmeasured = 2-exp(-(xideal-1)) but the exact form of the nonlinearity isn’t critical.) The measured spectra for concentrations from 0 to 3 is shown below, with concentration = 1 shown as the thick blue line. It is apparent that the shape changes as the concentration exceeds 1.


If the concentration is estimated using the ideal (concentration < 1) response, the estimate will fall below the actual value as the concentration passes 1, as shown below. If the spectral residuals were observed it would be apparent that there was a problem, but how to fix it?


If the ideal response for each concentration is estimated, then the difference between it and the observed response can be calculated, as shown in the top panel in the figure below. Because each difference spectra has a slightly different shape, the rank of this difference matrix is equal to the number of samples exhibiting non-linear behavior, which in this case is 20 (the samples with concentration 1.1 to 3). However, it is easy to get a basis for the nonlinear deviations using the Singular Value Decomposition (SVD). Furthermore, the singular values indicate that 93.7% of the residual sum of squares is captured in the first factor, and 98.6% is captured in the first two. The ideal response along with the first two basis vectors is shown lower panel.


When the CLS model is augmented with the two basis vectors, the prediction improves dramatically. The figure below shows the predicted concentration of the analyte as well as the “concentration” of the two additional basis vector factors. The correction added by the 1st nonlinear factor becomes quite large at high concentrations, whereas the contribution of the 2nd nonlinear factor remains relatively small. The prediction error in the concentration of the analyte is less than 1%.


In a future blog post we’ll explore some other aspects of CLS models.


The MathWorks released MATLAB R2014b (version 8.4) last week, and right on its heels we released PLS_Toolbox 7.9. R2014b has a number of improvements that MATLAB and PLS_Toolbox users will appreciate, specifically with graphics. The new MATLAB is more aesthetically pleasing to the eye, easier for the Color Vision Deficiency (CVD) challenged, and smoother due to better anti-aliasing. An example is shown below where the new CVD-friendly Parula color map is used to indicated the Q-residual values of the samples.


But the most significant changes in R2014b are really for people (like us) that program in MATLAB. For instance, TMW didn’t just change the look of the graphics, they actually changed the entire handle graphics system to be object oriented. They also added routines useful in big data applications, and improved their handling of date and time data. When you start the new MATLAB the command window greets you with this:

MATLAB R2014b Command Window at Startup

“Some existing code may need to be revised to work in this version of MATLAB.” That is something of an understatement. In fact, R2014b required the update of almost every interface from PLS_Toolbox 7.8. Revising our code to work with R2014b required hundreds of hours. But the good news for our users is that we were ready with PLS_Toolbox 7.9 when R2014b was released AND, as always, we made our code work with previous versions of MATLAB (back to R2008a). This, of course, is the significant difference between a supported commercial product and freeware. Not only do you get new features regularly, but you can rely on it being supported as operating systems and platforms change.

So if you look at the Version 7.9 Release Notes, you won’t see a lot of major changes. Instead, we took the time to assure compatibility with R2014b and made many minor changes to improve usability and stability.

The new MATLAB will allow our command-line and scripting users to do their science more efficiently and present their result more elegantly. These improvements will benefit us as well, and will ultimately translate into continued improvement in PLS_Toolbox and Solo.


On New Year’s day 2014 Eigenvector Research, Inc. (EVRI) celebrated its 19th birthday and began its 20th year. The momentum that carried us into 2013 built throughout the year and resulted in our largest year-over-year software sales increase since 2007. Our best three software sales months ever have all been within the last five months. Clearly our partnering with analytical instrument makers and software integrators plus our tools for putting models on-line are striking a responsive chord with users.

The consulting side of our business also continues to be very busy as we assist our clients to develop analytical methods in a wide variety of applications including medical, pharmaceutical, homeland security (threat detection), agriculture, food supplements, energy production and more.

The third leg of our business, chemometrics training, continued unabated as we taught on-site courses for government and industry, courses at conferences and held the 8th edition of our popular Eigenvector University (EigenU). We enter 2014 firing on all cylinders!

Major additions to PLS_Toolbox and Solo in 2013 included the Model Optimizer, Hierarchical Model Builder, a new Artificial Neural Network (ANN) tool, and several new file importers. We will soon release an additional ANN option along with new tools for instrument standardization/calibration transfer. Also on the horizon, a major new release of Solo_Predictor will include an enhanced web interface option and additional instrument control and scripting options.

2014 includes a busy schedule with conferences, talks, conference exhibits and short courses. Below is a listing of where you’ll be able to find us:

  • January 21-24, IFPAC, Arlington, VA. BMW to present “Mixed Hierarchical Models for the Process Environment” and “A Multivariate Calibration Model Maintenance Road Map.”
  • March 2-6, Pittcon Chicago, IL. NBG and RTR will be at the EVRI exhibition booth.
  • April 27-May 2, EigenU 2014, 9th Annual Eigenvector University, Seattle, WA. Join the complete EVRI staff for 6 days of courses and events.
  • May 6-9, EuroPACT, Barcelona, Spain. BMW to give plenary address “Model Maintenance: the Unrecognized Cost in PAT and QbD” and a condensed version of our “Chemometrics without Equations” short course.
  • June 1-4, CMA4CH, Taormina, Italy. JMS to teach short course and talk TBD.
  • June 8-12, CAC-XIV, Richmond, VA. NBG and RB to teach “Advanced Preprocessing for Spectroscopic Applications” and “Alternative Modeling Methods in Chemometrics.”
  • August 2-8, IDRC, Chambersburg, PA. NBG to attend, talk TBD.
  • September 14-18, ICRM, Nijmegen, The Netherlands. NBG to give keynote “An Overview of Hyperspectral Image Analysis in Chemometrics.”
  • September 28-October 3, SciX 2014, Reno, NV. JMS Chemometrics Section Chair, talks and courses TBD.
  • November 10-13, EigenU Europe, Hillerød, Denmark. Courses led by BMW and Eigenvector Associate Rasmus Bro.
  • November 17-19, EAS 2014, Somerset, NJ. EVRI sponsor of Award for Achievements in Chemometrics. Courses and talks TBD.

We’re especially excited about this year’s Eigenvector University. This ninth edition of EigenU will include all our usual events (poster session, PowerUser Tips & Tricks, workshop dinner) plus five new short courses. Special guest Age Smilde will lead “Chemometrics in Metabolomics” and Rasmus Bro will present “Modeling Fluorescence EEM Data.” The other three new courses are “Calibration Model Maintenance,” “PLS_Toolbox Beyond the Interfaces” and “Getting PLS_Toolbox/Solo Models Online.” We expect EigenU 2014 to be an especially fun and fruitful learning experience.

We look forward to working with you in 2014!


26th Aug, 2013

Eigenvector Summit 2013

The Eigenvector staff is pretty spread out geographically. We have people in four cities in Washington, plus folks in Indiana and North Carolina. And then there’s our Associate Rasmus Bro in Denmark. So we don’t get together, other than virtually, very often. Generally we’re only together at Eigenvector University each spring in Seattle. But we’re so busy during that time that we don’t get to talk to each other much.

This July we remedied that by having our first ever Eigenvector Summit. We got together for a week in Manson on Lake Chelan where we discussed the future of Eigenvector’s efforts in software, training and consulting. Plus we had some great food and beverages, did a few tourist activities, visited some local hangouts, went out on the lake and, most importantly, enjoyed each other’s company.

Developers at EigenSummit 2013

The EVRI developers are pictured above. Left to right are R. Scott Koch, Randy Bishop, Bob Roginski, Barry Wise, Donal O’Sullivan, Jeremy Shaver and Neal Gallagher with Rasmus Bro reclining. For a few more pictures from the Summit, see Scott’s slide show.

As is typical when we get together, we never ran out of things to talk about and try. The guys were often around the dining room table with their laptops writing demo code or pointing each other at new journal references. Other times we were out by the lake just brainstorming. (Thanks, Bob, for taking the notes!) And while I won’t reveal what we came up with, we expect that Eigenvector Summit 2013 will pay dividends for our users, students and clients in the years that come.


Emil W. Ciurczak wrote a nice blog post for PharmaEvolution claiming that the weak link in QbD is a lack of adequately trained chemometricians. I enjoyed his article, Chemometrics: The Weak Link in QbD, and I agree wholeheartedly with the conclusion that “the need for correct and in-depth chemometrics training is necessary for a successful PAT program.” But I don’t see how this is reconciled with “The title and spirit of this brief educational brochure from one vendor are right on target.” Anything that is “For Dummies” almost surely cannot be “in-depth.” And while I applaud the vendor for coming up with the concept, (and I’d say I wish we’d thought of it except we did think of it and the result is our CWE-Chemometrics without Equations courses) there isn’t much real information in the brochure. For instance Chapter 4 on classification is just 3 pages, one of which is just a list of applications.

The fact that chemometrics remains the weak link iin QbD is disappointing but it certainly isn’t for lack of effort on our part. We’ve taught hundreds of classes and thousands of students but a fairly small fraction of those are from pharma. From our business point-of-view, but also as a consumer of pharma products, I’d be happy to see more effort go into developing staff with chemometrics expertise. I do appreciate that it is, for many people, rather challenging subject material. We have put great effort into making chemometrics accessible. But I resist the urge to dumb it down too much.

An often used analogy is that most people don’t know how their mobile phone works but they are still able to use it, and chemometrics should be just as easy. But, unlike cell phones, chemometric tools aren’t being used by consumers, they are part of the process for producing things like mobile phones. And drugs. As such, their use and misuse has consequences. And while I’m happy to introduce people to chemometrics with our CWE courses and think attendees gain a useful level of proficiency and understanding of the techniques involved, I would prefer that those involved in QbD and pharma manufacturing acquire a deeper level of mastery. (To this end we provide our Eigenvector University courses, the next instance of which is EigenU Europe this October.) In order to understand a system’s limitations and how it can fail, you really need to understand how it works. You’re not going to get that at the “Dummies” level.

I very much appreciate Emil’s continued efforts to enlighten pharma as to the critical role of chemometrics. But, like me, does he sometimes feel as though he is pushing a string? I was struck by the registration process for PharmaEvolution website when you had to select your company’s business. There was a very long list of possibilities, but the closest thing to what we do was the very generic selection “engineering.” I still get the feeling that many in pharma (and certainly in some other industries) think of chemometrics as something that you do AFTER you’ve decided everything else and have started to take data. Thank-you, Emil, for your efforts to make it the integral part of the system that it must become.


Each year at Eigenvector University we host a PLS_Toolbox/Solo User Poster Session and User Group Meeting. This is a fun event as it gives our users a chance to show what they have been doing with our tools. It is also a chance for them to relax, have a beverage, and give us input on upcoming versions of our software.

This year’s EigenU Poster Session will be Tuesday, May 14, at 6:00 at the Washington Athletic Club. PLS_Toolbox and Solo users will showcase their own chemometric achievements and share results and remaining problems with other users and the EigenU instructors. We’ll also have a brief User Group Meeting where attendees can see what’s in the future of Eigenvector software development and give their input and feature requests.

The poster session and user group meeting will include complimentary beverages and hors d’oevres. This year’s Best Poster grand-prize is an Apple iPad mini (32GB). The runner up will receive an Apple iPod nano. Judging will be done by the EVRI staff. Attendance at the poster session is free and open to all EigenU attendees and Eigenvector software users. You need not attend classes at EigenU to come!

If you would like to present your work, please send a title and brief abstract to Please be sure to describe how PLS_Toolbox, Solo or our other software products were used in the work.

See you there!


We have developed a collaboration with FOSS where we teach an open chemometrics course at their World Headquarters in Hillerød, Denmark, each fall. During our last course, the FOSS folks asked if I would do an interview for their website. The website is a resource for Feed industry professionals who work with Near Infrared (NIR) spectroscopy.

The interview is a little bit slanted towards NIR users but contains some good general information about our courses, consulting and general chemometrics philosophy. In the video I answer the following questions about chemometrics and our courses:

  • Whom are your chemometrics courses for?
  • How much of an NIR solution is based on chemometrics?
  • What does the average NIR user need to know about chemometrics?
  • What are the trends in chemometrics today?
  • How will your course attendees use their training in chemometrics?

Our next big training event is Eigenvector University 2013. This 8th Annual EigenU runs May 12-17 in Seattle. We have a number of other training opportunities in 2013, including courses in the UK, Spain and France. See our schedule for details. We also plan to be back at FOSS for EigenU Europe this October.


One of the challenges of writing software that works with MATLAB is accommodating an array of versions. For better or worse, not everybody updates their MATLAB regularly. So we have to make our PLS_Toolbox and other toolboxes work with a fairly wide distribution of MATLABs.

To give you some idea of what our developers are up against, the plot below shows the distribution of MATLAB versions among our users for each of the last three years. (Click on the plot to get a much larger .pdf version.)

While the most common version in use at any one time tends to be one of the latest two or three releases, it never peaks at more than 20% of our users. And there are LOTS of users with older versions of MATLAB. Note that the plot goes back ten years to 2003! In 2010, we still had 12% of our users with MATLAB versions from 2005 or earlier. It was only after that dropped to less than 5% that we stopped supporting MATLAB 6.5 and 7.0.1 in our new releases. As shown in our release notes, we currently support MATLAB 7.0.4 (from early 2005) through the current MATLAB 8.0 (R2012b). And with our latest minor update (PLS_Toolbox 7.0.3) we’re ready for R2013a, so you’ll be set when it comes out.

But it is a balancing act. We don’t want to force users to upgrade their MATLAB. We understand that an older version of MATLAB works perfectly well for many users. But often we can’t take advantage of newer MATLAB features until we cut old versions loose. As an example, it would be much easier for our developers to use the newer format for coding objects (such as our DataSet Object) that became available in MATLAB 2008a. Until recently, however, 10% of our users were still working with MATLAB 2007b or older.

Our Chief of Technology Development Jeremy M. Shaver notes: Moving users to later versions of MATLAB allows us to utilize better graphical interface tools (making our interfaces easier to use and more powerful), modern hardware architecture (allowing faster processing and better memory management), and other new programming functionality (making the code easier for us to support and for our power-users to understand). Plus, having fewer MATLAB versions to support means we have fewer “special cases” to support in the code. We balance this against our user’s inconvenience and cost in order to achieve the best overall result for our customers!

Well said, Jeremy!


2012 is in the books and it was another good year for EVRI. Once again our year-over-year software sales were up. The end of the year was especially good. November was our best software sales month ever until December, which was better still! We move into 2013, our 19th year, with a renewed sense of momentum.

In 2013 we’ll continue to build on the unique features of our software. As always, improvements and additions are driven by the combination of customer requests, our own needs for consulting projects, and interesting developments in the field of chemometrics. We will continue to release two major updates a year (spring and fall) and minor updates as required. Updates are free to users with active maintenance agreements. Software support continues to be a top priority, and we have extended the hours of coverage for our e-mail based, now 6am to midnight Eastern Time, Monday-Friday, (with additional checks on weekends).

Software improvements planned for 2013 include:

  • Methods to help automate the model building process in PLS_Toolbox and Solo
  • Automatic evaluation of different model types and preprocessing schemes
  • Improvements to Solo_Predictor for on-line analysis, including expansion of the number of ways to communicate with other software and instruments
  • Compatibility with more instruments from more manufacturers, more Technology Partners

Our training efforts will continue unabated in 2013, with our 8th annual Eigenvector University planned for May 12-17. You’ll also find our courses at numerous conference sites. Our EigenU Online classes will be expanded, with classes on Clustering and Classification and Multivariate Image Analysis coming this Spring. And of course we’ll continue to add to our bank of EigenGuide Videos demonstrating use of our software.

EVRI’s staff of consultants and software developers, with a combined 100+ man-years of chemometrics experience, will continue assisting clients both domestically and internationally.

We look forward to working with you in 2013!


I got an email from a prospective user of our software the other day that really set me back. Paraphrasing a bit here, it was “Are there any unique features of your PLS algorithm/diagnostics?” The problem with questions like this one is that I never know where to start. But here is what I wrote.

As for “unique features of your pls algorithm,” well, there are numerous ways to calculate a PLS model, but they all pretty much arrive at the same result (which is good). If you’d like to learn more about PLS algorithms and their accuracy, I suggest you have a look at a series of blog posts I did on the subject. See:

Accuracy of PLS Algorithms
Re-orthogonalization of PLS Algorithms
One Last Time on Accuracy of PLS Algorithms
Speed of PLS Algorithms

As to diagnostics, most of the packages use pretty much the same diagnostics, though sometimes they call them by different names. Usually there is a sample distance metric (e.g. T2) and some sort of residual (e.g. Q).

But maybe what you are really looking for is what makes our software unique, rather than our specific PLS algorithm. We have two major packages for chemometrics. The first is our MATLAB-based PLS_Toolbox, the second is our stand-alone product Solo, which is essentially the compiled version of PLS_Toolbox. The two packages provide identical interfaces and share the same model and data formats. The advantage of PLS_Toolbox is that, because it works within the MATLAB environment, it can be run from the command line and functions from it can be incorporated into other analyses. The advantage of Solo is that you don’t have to have MATLAB.

So right off the bat, a unique feature of our software is that there are completely compatible solutions for working with or without MATLAB. And both of these solutions are available on all platforms, including Windows, Mac OSX and Linux. That is unique.

PLS_Toolbox and Solo have the widest available array of analysis methods. This includes PLS and PCA of course, but also PCR, MLR, MCR, PARAFAC, N-PLS, PLS-DA, SIMCA, SVM, KNN, CLS, LWR, MPCA, Cluster Analysis and Batch Maturity. Plus they have a large number of auxiliary tools for Instrument Standardization, Data Transformation, Dynamic Modeling, Sample Selection, Trend Analysis, Correlation Spectroscopy and Design of Experiments. And numerous tools for variable selection including Genetic Algortihm, iPLS and Stepwise MLR. Plus diagnostic methods such as VIP and Selectivity Ratio. The collection of all of these analysis methods and auxiliary functions with one interface is unique.

PLS_Toolbox and Solo can be extended for use with Multivariate Images with MIA_Toolbox and Solo+MIA. The ability to apply such a wide array of multivariate analysis techniques to images is unique. There is also an add-on for the patented Extended Multiplicative Scatter Correction, EMSC_Toolbox. If not completely unique, this method for preprocessing data from highly scattering samples is not widely available.

For on-line application there is our Solo_Predictor and Model_Exporter. Solo_Predictor can be used with any model generated by PLS_Toolbox/Solo and can communicate via TCP/IP sockets, ActiveX, .NET, timed action or wait-for-file. Model_Exporter translates PLS_Toolbox/Solo models into mathematical formulas that can be compiled into other languages. Model_Exporter’s XML output can be parsed for execution in .NET (C#). Additional output formats include MATLAB .m file (compatible with older versions of MATLAB and OCTAVE, plus LabView, Symbion and Tcl). This wide array of on-line options is unique.

Beyond that, PLS_Toolbox and Solo are also extremely flexible tools and include the widest array of data preprocessing methods with user-specified ordering, ability to add user-specified method, and customizable favorites settings.

And finally, price. PLS_Toolbox is only $1395 for industrial users, $395 for academic. Solo is $2195/$695. The price/performance ratio of these products is most certainly unique.

If you have any questions about the specific functionality of our software, please write me.


3rd Dec, 2012

Bruce Kowalski

I was sitting in a cubicle in the United Club at Chicago O’Hare when I learned of Bruce Kowalski’s passing. The news was not unexpected, but it was still tough, in part due to my whereabouts in a busy place but with no friends or family. Additional memories factored in: I had also been in a United Club cubical when I heard of my father’s death. But this was especially ironic because Bruce was the reason I was there in the first place, on my way to Lille for ChemomeTRIcS in Time-Resolved and Imaging Spectroscopy.

I met Bruce on what turned out to be the most pivotal day of my life, October 1, 1985. My existence divides between everything that came before that day and everything that came after. It was my first day of graduate school, the day I started learning about chemometrics, and also the day I met my friend and business partner Neal Gallagher. Bruce introduced me to the discipline that I became immediately enamored with and have spent my working life on. Bruce also introduced me to his then Post-Doc Dave Veltkamp and his wife Diane who in turn introduced me to my wife, Jill. This accounts for pretty much everything else in my present life.

An “idea guy” who’s enthusiasm was infectious, Bruce’s achievements include co-founding, with Svante Wold, the field of chemometrics, and co-founding, with Jim Callis, the Center for Process Analytical Chemistry. The full breadth of Kowalski’s influence is really too big to capture in this small space and couldn’t be done without substantial research. But as an example of his influence in one area I submit a graphic prepared by Pieter Kroonenberg and presented at TRICAP 2000. The “Kowalski Web” demonstrates that, for the field of Multi-way Analyisis in Chemistry and as far as it has become connected to Multi-way Analysis in Psychology, all the connections lead to Bruce, the center of the web. If this graphic were updated today Bruce would still be at the center, but it would be much larger!

I could go on and on about Bruce’s influence. At least two software companies exist today because of him, Infometrix and Eigenvector. Chemometrics has enabled the development of countless sensor systems and greatly expanded applications of spectroscopy, especially NIR. The methodologies promoted by Bruce have become so pervasive that quantifying their impact would be a very large exercise.

Bruce taught me lots of things, both chemometricly and otherwise. The chemometric stuff is (I hope) obvious, so I’ll let that speak for itself. Beyond that, though, Bruce taught me to think big, that a good idea can’t be stopped, that it is important what you name things, to be magnanimous, and to spread the credit around. Bruce was always really good at talking up the people that worked for him. He found the best in people, let them know it, and then let other people know about it too. I’ve benefited greatly from that, as have many others. I hope that I’m as good with my staff as Bruce always was with his. And I strive to be as forward-looking, positive and fun to be around.

Farewell, Bruce. A little bit of you will live on in each of very many of us. You will remain in our thoughts and prayers.


Eigenvector’s Chief of Technology Development Dr. Jeremy Shaver is getting ready to head off to the Eastern Analytical Symposium (EAS). He’ll be busy on Sunday and Monday assisting Eigenvector Associate Dr. Don Dahlberg with Chemometrics without Equations (CWE). As I wrote previously, this year the popular CWE is being extended by a day to cover advanced data preprocessing. Jeremy will be demonstrating the methods using the recently released PLS_Toolbox/Solo 7.0. If you’d like to attend, there is still time to register through the conference web site!

Jeremy will also represent EVRI at the session honoring Professor Dr. Lutgarde Buydens of Radboud University Nijmegen for Outstanding Achievements in Chemometrics. The award is, once again, sponsored by Eigenvector Research. The award session, chaired by University of Barcelona’s Dr. Anna de Juan, will start Monday morning at 9:00am.

You might also find Dr. Shaver at the Cobalt Light Systems Ltd booth. Cobalt, one of EVRI’s Technology Partners, develops tools for non-invasive analysis. Their TRS100 pharmaceutical analysis instrument utilizes our Solo software for chemometric modeling. Jeremy will be there to advise users on how to best calibrate the system for their particular needs.

Of course, if you can catch him, Jeremy would be happy to talk to anyone interested in EVRI’s software offerings! He’s the Eigenvectorian most intimately familiar with our products and their features and capabilities. Drop Dr. Shaver an email if you’d like to meet him at EAS.

Have a good week!


New versions of our MATLAB-based PLS_Toolbox and MIA_Toolbox were released earlier this month, along with updates to our stand-alone packages Solo and Solo+MIA. PLS_Toolbox and its derivatives, Solo and Solo+MIA, are now in version 7.0, while MIA_Toolbox is in version 2.8. As can be seen in the release notes, the list of enhancements and additions is long (as usual!).

Many of the new features are demonstrated in the new EigenGuide video, “What’s New in Version 7.0.” The video illustrates the use of:

  • additional information in the Analysis interface, such as error of cross-validation
  • interfaces for splitting data sets into calibration and validation sets
  • tools for visualizing the difference between samples for both their Q residuals and T2 contributions
  • simplified control of plot attributes
  • readily available class statistics
  • automated peak finding
  • tools for finding specific samples and variables based on logical operators

Of particular note in this release is the expansion of the Batch Process Modeling tools. The Batch Processor tool readies data sets for modeling by Summary PCA, Batch Maturity, MPCA, and several PARAFAC variants. It then pushes the data sets into the Analysis tool where the models are developed. To see the Batch Processor and Analysis in action, watch the video. The combination of the Batch Processor and methods supported in the Analysis interface allows modelers to follow most of the pathways outlined in my TRICAP 2012 talk, “Getting to Multiway: A Roadmap for Batch Process Data.”

This release reaffirms EVRI’s commitment to continuous software improvement – it completes our fifth year of semiannual major releases. The best chemometrics software just keeps getting better!


25th Oct, 2012

Thanks to FOSS!

Rasmus Bro and I wrapped up another short course today, this time at FOSS World Headquarters in Hillerød, DENMARK. We’d like to thank our students, another great group, and especially our host at FOSS, Lars Nørgaard, Senior Manager of FOSS Team Chemometric Development. Lars makes it easy for us to hold this course as the meal and coffee service arrangements are all done through the excellent FOSS Canteen, and they also provide a very comfortable classroom.

This was our second year at FOSS, and we’re planning on making it a regular thing, our Eigenvector University Europe. Next year we’ll be in FOSS’ new “Innovation Centre,” due to be completed this winter. We’re certain it will provide a great venue for EigenU Europe, 2013. Exact dates are TBD, but we’re looking once again at the last half of October.

Thanks again to the kind folks at FOSS!


The popular Chemometrics without Equations (CWE) series will be extended at the Eastern Analytical Symposium (EAS) this year with the addition of a second day, CWE II. While the original CWE focuses on the basics of Principal Components Analysis (PCA) and Partial Least Squares (PLS) regression, CWE II will explore advanced data preprocessing methods and mixture analysis.

I have been known to say that the secret to getting good regression models is what you do before the data hits the modeling algorithm. Preprocessing methods attempt to remove extraneous variance so that the variance of interest can be more easily modeled. When done correctly, preprocessing can greatly improve model performance. CWE II covers the whys and hows of data preprocessing with examples from several methods.

Also covered in CWE II, Mixture analysis techniques, such as Multivariate Curve Resolution (MCR) aka Self-Modeling Mixture Analysis (SMMA), can elucidate the true underlying physical roots of the data, e.g. pure component spectra and chemical concentrations. As such, these methods can lead to better fundamental understanding of the systems involved.

The courses will be led by Dr. Don Dahlberg, Emeritus Professor at Lebanon Valley College. Don will be assisted by EVRI’s Chief of Technology Development Dr. Jeremy M. Shaver.

The courses will be held Sunday and Monday, November 11 and 12, at the Holiday Inn in Somerset, NJ. Registration is handled by EAS. Complete course information can be found in EAS’s Short Course Schedule and Description.