SIAM SDM'11 Data Mining Challenge: Prediction of   Biological Properties of Molecules from Chemical   Structure  (PDF)
The problem was to classify chemical molecules into substrates and non-substrates of the CYP 2C19 isoform of the cytochrome P450 enzyme in human. Predicting whether or not a particular chemical will be metabolized by 2C19, as well as other major isoforms of CYP P450, is of primary importance to the pharmaceutical industry. The challenge, which was entirely based on numerical data, only.
We submitted a top three solution using KnowledgeMiner Insights right out of the box.

Modelling and Prediction of Toxicity of Environ-
mental Pollutants (PDF)
Introduction, problem description, and concept outline for building Quantitative Structure-Activity Relationship (QSAR) models of ecotoxicological compounds like pesticides to reduce animal tests for dossier preparation for regulatory purposes.

Selected Publications
A brief collection of publications by Insights users related to application of GMDH self-organizing data mining technologies in various domains.

NASA, Boeing, MIT, Columbia, Notre Dame, Mobil Oil, Pfizer, Merck, Dean & Company, and many other corporations, universities, research institutes and individuals around the world.

"Insights is the most advanced implementation of the GMDH approach now. It uses the inductive method, which is different from deductive techniques used commonly for modeling on principle. Many important successful results were received using this tool recently. They show the advantage of it over analogous well-known software."
— Prof. Alexey G. Ivakhnenko, author of the GMDH approach

"Insights' data mining and self-specifying model capabilities help us develop more robust and useful tools for ourselves and our clients. We have built regional and national macroeconomic models, securities trading tools, event study approaches, and microeconomic market characterizations. Specifically, Insights is at its most useful when the natural systems are complex and the a priori structure of the equations are unknown. In such settings Insights' data mining and self-organizing capabilities are extremely helpful."
— Dr. Randall Pozdena, ECONorthwest, an economics and finance consultancy

"Insights is the only product that I have found that makes it easy to try non-standard equation formats on a data set. Many standard regression tools are as easy, but they limit you to a small set of potential relationships. KnowledgeMiner combines spreadsheet-like set up with an algorithm that doesn't "over fit" the model. Also, the output is in a readily usable format (e.g. not C++ code)."
— Ware Adams, Dean & Company, a strategy consulting firm

"I like Insights because its algorithm does not make any assumtions on the underlying data; well, at least not during the initial model-building phase. I also like the fact that it generates sets of equations that the user can review with detailed understanding of the interactions and dependencies of each variable. Also, the algorithm(s) behave surprising well under extreme conditions for certain complex dynamical systems. Congratulations for your excellent work."
— Alexis Pobedonostzeff, Pfizer Inc. Director, Health Care Issues Analysis & Management

"I'm a physicist by training, working as a radar engineer on some cutting edge target recognition/classification technologies. I am now using Insights to circumvent all the past pattern recognition algorithms which have been years (and millions of $) in development by the armed forces. Although I am just now starting to use Insights in this application, my initial indications are that Insights is providing a more robust, complete and more accurate classification capability than any of the previously used algorithms, and with comparatively no effort on my part"
— Herb, Vista Technologies, Inc.

"Your results are very impressive... If it's not a joke, you hold of the nitroglycerine! I have never heard about this kind of research anywhere in the press. Best wishes for you all."
— Maxime, Institut de Physique du Globe de Paris

"I have purchased your program Insights and have had some time to use it. My research is in artificial intelligence applications in clinical medicine at the University of Western Ontario in London, Canada. I have so far used backward error propagation and probalistic ANNs for outcomes based research. I also have some experience with fuzzy decision theory and expert systems. Your program looks interesting and has some advantages over my current modelling software. ... I wish to congratulate you on your very promising software."
— Wayne, Associate Professor of Medicine, Division of Cardiology, University of Western Ontario, London, Canada

"Lovely maths and algorithms. Nice and simple product. Feel that it can significantly assist me. Looking forward to understanding it better to put it to real use."
— Dr. Conrad Mackenzie, Australia

"I am Head of Computing and Information Technology at Katikati College, a high school in New Zealand. A couple of weeks ago I attended a big AppleFest at Rotorua with attendees from about 120 different schools. One key speaker delivered a workshop where he named the two big Mac products of 1998, one was Myrmidon and the other one (of course) was KnowledgeMiner. I have downloaded the demo and it appeals to me because of my interest in AI in general and neural nets in particular."
— John, Katikati College, New Zealand

"I would just like to congratulate you on this program on the behalf of Roger Bradbury who did some work on GMDH back in 1988 (Green, D. G., Reichelt, R. E., and Bradbury, R. H. (1988) Statistical Behaviour of the GMDH algorithm. Biometrics 44: 49-69). He is happy that there is a modern version of it - of which we will definitely be purchasing."
— Belinda, Bureau of Resource Sciences, Australia

"I believe that tools like this are definitely the start of something very big in getting a handle on mountains of information."
— Douglas, Dartmouth Medical School

"The Alpine skiing and Athletic French Federation have contacted my laboratory to build a profile of their elite athletes. In this case, KnowledgeMiner helped me save a lot of time and gave me models on the most important variables, and pointed out the less relevant."
— Fabrice Viale, Doctoral thesis student Laboratoire de Physiologie, Faculte de Medecine, France