What are the system requirements?
TheGmax runs in all Windows operating systems. It functions on both 64bit and 32 bit systems and is available as a stand-alone or network installation.
- Predictive data-mining & goal optimization for Business, Life sciences and Genomics.
- Reduces high dimensional problems to low dimensional solutions.
- Simplifying complexity.
- Say goodbye to regression, you are unlikely to need it again.
TheGmax runs in all Windows operating systems. It functions on both 64bit and 32 bit systems and is available as a stand-alone or network installation.
All popular application formats such as SPSS, SAS, Excel, Statistica, QuatroPro, Matlab, Lotus etc. and any ODBC compliant database. New file formats can usually be added on request.
There is a natural limit to number of cells that TheGmax can absorb, due to its 32bit architecture. The practical limit is about 66 million cells (= Rows x Columns), which can be raised to 99 million cells by closing down all non essential programs. If the incoming data exceeds your computer's available memory then TheGmax will offer to draw a representative sample of observations. Alternatively you may use the built-in custom sampling feature to draw a sample of your own design.
TheGmax accepts both text (categorical) and numeric (ordinal and continuous) data. Missing data is tolerated. It is assumed that the data can be cast as a rectangular grid such that the columns are variables and rows are observations. If the reverse is true then TheGmax can rotate the file.
Yes. TheGmax can model any categorical variable. For purely cosmetic reasons it works best if the number of categories to be separated is 25 or less. However, there is no actual restriction on the number of categories.
TheGmax should be an essential tool in every analysts armoury. It may not be a panacea but it excels at non-parametric optimization and competes strongly with all known paradigms. The closest competition comes from Random Forests which regularly match TheGmax predictions. However, unlike TheGmax, a Random Forest is not a single expression but a collection of models that offer differing results. TheGmax is still the only product that can directly optimize non-parametric and rank base measures of association in a single step.
TheGmax is leased, not sold. There is no front loading. Non-profit organizations usually attract a very large discount. Installations are tailored to meet the needs of the user, which can be anything from a single PC to a 500 seat network spread across a continent.
GP is an Automated Invention Process that was authored by John Koza of Stanford University in 1990. Many websites describe GP and we encourage you to dive in. GP is unique because it is not a computer program per se, but rather the mechanism that enables software to breed, via natural selection methods, and ultimately evolve into algorithms that are perfectly adapted to their intended purpose. This makes GP a perfect choice for non-parametric goal optimizations. No other paradigm can do this.
If an independent nominal variable has many categories, such as zip codes, then TheGmax groups together categories that appear to exhibit similar characteristics. These groups or 'bins' provide a shorthand reference to large numbers of categories that would otherwise lead to substantial bloating of the model trees, which in turn would slow the evolutionary process considerably.
The method, the math, the software and the look-and feel were invented, authored and designed in 1990 by London based statistician, Michael Sinclair, to serve as an analytical tool for the research consultancy Frost-Sinclair. The software was named superNATURAL and later abbreviated to sNAT. Contributions and feedback from several users during the 90s resulted in the product's re-branding as TheGmax in 2001. Since that time TheGmax has matured hugely and now exhibits astonishing power in terms of both scale and scope.
There are still a few cloned imitations that use a pirated version of our 1998 software. They have limited ability, weak performance and lack support. Despite some visual similarities, they are easily spotted because they do not display our logo "TheGmax". It is of course flattering to have ones software pirated; it is testament to having created something greatly desirable and of course it provides free advertising for the real thing, TheGmax.