Under Same (Old) Management

That’s not a headline you see very often. Usually it’s “Under New Management.” But here at Eigenvector Research we’re proud of our stability. I wrote the first version of our MATLAB-based PLS_Toolbox while I was in graduate school thirty-one years ago. I still oversee its development along with our other software products.

In 1990 Partial Least Squares (PLS) regression was still fairly novel. PLS_Toolbox 1.0 included it, of course, along with a non-linear version of PLS and a number of tools for Multivariate Statistical Process Control (MSPC) including Principal Components Analysis (PCA). The goal then, as it is now, was to bring new multivariate modeling methods to users in a timely fashion and in a consistent and easy to use package.

Neal B. Gallagher joined me in 1995 to form Eigenvector Research, Inc. He has been contributing to PLS_Toolbox development for almost 27 years now, along with consulting and teaching chemometrics, (i.e. chemical data science). Our senior software developers R. Scott Koch, Bob Roginski and Donal O’Sullivan have been with us for a combined 45 years (18, 15 and 12 respectively). That continuity is one reason why our helpdesk is actually so helpful. When you contact helpdesk with a question or problem we can generally get you in touch with the staff involved in writing the original code.

To assure that continuity going forward we’ve brought some younger developers on board including Lyle Lawrence and Sean Roginski. (Lyle was still sleeping in a crib and Sean wasn’t born yet when PLS_Toolbox first came out-ha!) Both have taken deep dives into our code and have been instrumental in the recent evolution of our software. Primarily on the consulting side of EVRI, Manny Palacios brings his youthful energy and extensive experience to our clients’ data science challenges.

Over the years we have developed and refined PLS_Toolbox along with our standalone software Solo, adding many, many new routines while advancing usability. Currently we are completing the process of integrating new methods from the Python libraries scikit-learn and TensorFlow into the soon to be released PLS_Toolbox/Solo 9.0. So when we bring you new methods, like Deep Learning Artificial Neural Networks (ANNDL, shown above) or Uniform Manifold Approximation and Projection (UMAP, below) you can be sure that they are implemented, tested, supported and presented in the way that you’ve come to expect in our software. They have the same preprocessing, true cross-validation, graphical data editing, plotting features, etc. as our other methods.

Now, 25+ years in, we’re moving forward with the same vision we’ve had from the beginning: bring new modeling methods to the people that own the data in a consistent straightforward package. This same old management is working to assure that far into the future!

BMW