Statistical and Multivariate R&D

Over the last 3+ decades of performing statistical and multivariate analysis, we have gained a great deal of experience in this area. Some of the approaches used here and not readily available elsewhere include the following:

· HVA (Hyperplanar Vector Analysis - unique to GXStat): A system to decompose a multivariate data set into end-members and proportions, regardless of whether or not the end-members are contained in the data or not, where the original data may not be defined by constant row sum (Polytopic Vector Analysis - PVA - requires this).  The composition of the end-members are presented in the original metric and there are tools used to define the proper number of end-members.

· FCM (Fuzzy c-Means): A program based on the article by Bezdek, Ehrlich and Full, Computers & Geosciences 10(2-3):191-203 (1983).  An alternative to hard clustering that minimized the effect of extreme data - a common component of large, geologic data sets.

· FNV (Fuzzy N-Varieties): A program based on the book by Bezdek, Pattern Recognition with Fuzzy Objective Function Algorithms, Kluwer Academic Publishers Norwell, MA (1981).  This program addresses the issues with clustering about cluster center points and allows for clusters to be defined by lines, line segments, planes, hyperplanes and so forth in addition to cluster centers. This program is useful the decomposition of big data into more meaningful sub-spaces.

Given GXStat's broad experience not only creating and developing several of these programs, performing similar research and development to solve the customer's problem is a service we offer. GXStat's researchers have performed this service for both commercial and Department of Defense entities.

For more detailed information on GXStat's Statistical and Multivariate R&D services, contact us.