Symbolic data

Publication date 18 November 2001

categorical variables data matrix interval variables multi-valued variables quantitative variables set-valued variables symbolic data arrays symbolic variables

Mathematics Subject Classification ID

Review text

The author introduces the concept of symbolic data and defines various types of symbolic data. In his paper on ``The classical data situation, see the preceding review, Zbl 0978.62001, he described the classical data analysis paradigm: a rectangular data matrix \(X=\{x_{kj}\}_{k=\overline{1,n}}^{j=\overline{1,p}}\) defines a relation between the set \(\Omega=\{1,\dots,n\}\) of \(n\) individuals or elementary objects and a series of variables \(Y_1,\dots,Y_p\) defined on \(\Omega\), where each variable \(Y_j\) realizes one single category \(Y_j(k)\) from a range \(\mathcal Y_j\) such that each cell \((k,j)\) of the data matrix contains one single value \(x_{kj}\) only. A symbolic variable \(Y\) with domain \(\mathcal Y\) is a mapping \(E\to{\mathcal B}\) defined on a set \(E\) of statistical entries (individuals, classes, objects, etc.) and which takes its values in a range \({\mathcal B}\) of nonempty subsets \(U\) of the set \(\mathcal Y\). Depending on the specification of \(\mathcal B\) in terms of \(\mathcal Y\), symbolic variables can be classified according to the following scheme: (a) if \({\mathcal B}={\mathcal Y}\), we have the classical single-valued variable case; (b) \(Y\) is set-valued if \(Y(a)\subset\mathcal Y\) is a subset of \(\mathcal Y\) for all \(a\in E\); (c) \(Y\) is an interval variable if, for all \(a\in E\), \(Y(a)=[\alpha,\beta]\) is an interval of \(\mathcal Y\); (d) \(Y\) is a multi-valued variable (categorical or quantitative) if it is set-valued with \(Y(a)\subset\mathcal Y\) and \(|Y(a)|<\infty\) for all \(a\in E\); (e) \(Y\) is a modal variable with domain \(\mathcal Y\) if, for each \(a\in E\), \(Y(a)\) is a non-negative measure on \(\mathcal Y\), typically a frequency distribution, a probability distribution or a weighting. Together with the classical (single-valued) data types, these are the data types considered in the Symbolic Official Data analysis System (SODAS) project and termed symbolic data types. In contrast to a classical rectangular data matrix introduced in the author's paper cited above, the determination, sampling, observation or recording of symbolic data leads to a symbolic data array which is the basis of all symbolic data analysis methods. For the entire collection see [Zbl 1039.62501].

Recommendations

Cited in

(13)

Describes a project that uses

Uses Software

SODAS

This page was built for publication: Symbolic data

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2756624)