Statistical data analysis classification function

Modeling by using the structure contained within a data set is a problem-oriented process, and as such it is fundamentally opposed to notions of a static mode of group formation. In the recent archaeometric literature, it was proposed that, as a convenient method of communication between researchers, the classification functions derived from discriminant analysis could be transmitted rather than the actual data (11) Such a proposal seems to be insensitive to problem orientation, ceramic processes, or statistical influence. 

General entries concerning statistical indices, regression parameters, classification parameters, similarity/diversity were completely rewritten trying to give an exhaustive view of the functions used to characterize data, modeling, and similarity/ diversity analysis. For example, more than 50 distance and similarity functions have been reported. 

The adaptive least squares (ALS) method [396, 585 — 588] is a modification of discriminant analysis which separates several activity classes e.g. data ordered by a rating score) by a single discriminant function. The method has been compared with ordinary regression analysis, linear discriminant analysis, and other multivariate statistical approaches in most cases the ALS approach was found to be superior to categorize any numbers of classes of ordered data. ORMUCS (ordered multicate-gorial classification using simplex technique) [589] is an ALS-related approach which... 

Using several modifications discriminant analysis and related statistical classification methods have been widely and successfidly applied in the QSAR field by several workers [75-89, 91-104]. In our work we have always preferred to use discriminant analysis (or another adequate classification method) for data of low or unknown precision rather than to force such data into a regression model. A discriminant function (in multi-class problems, the first discriminant function) can often be interpreted analogously to a Hansch equation in mechanistic terms. 

Regression analysis (black box models) i.e., statistical methods that permit the approximation of functions and the classification of data using non-parametric methods (application specific). 

---