RNTI

MODULAD
The symbolic data analysis paradigm, discriminant discretization and financial application
In HDSDA 2013, vol. RNTI-E-25, pp.1-14
Abstract
The variability inside classes of individuals, categories (defined by a categorical variable) or concepts (defined by an intent and an extent, like species for example), is expressed by the use of intervals, histograms, distributions, sequences of weighted values and the like. In this way we obtain new kinds of data called "symbolic". The aim of "Symbolic Data Analysis" (SDA) is to study and extract new knowledge from these new kinds of data by an extension of Statistics and Data Mining to symbolic data.We show that SDA is a new paradigm opened to a vast field of research and applications. Then, we give a way for obtaining discriminate symbolic descriptions by an original discretisation method, which is illustrated by a financial application.