Symbolic approaches for three-way data
Let data be observed at \(T\) time points. For each date \(t\) we have a data array \(\underline{X}=(\xi_{kjt})_{n_t\times p_t}\), where each entry \(\xi_{kjt}\) is a symbolic data record such that the rows \(k\) may correspond to \(n_t\) (individual) symbolic objects and the \(p_t\) columns \(j\) correspond to symbolic variables such as quantitative or qualitative variables, multi-valued ones, interval-type or modal variables. The investigation of such data requires to adapt and generalize classical and symbolic data analysis methods to the case of time series. The study of classical two-way data arrays has been extended to three-way data arrays with the third dimension being usually time or space. In this chapter a summary of recent methods and developments on this topic is presented as well as their links to the symbolic approach and their realization through the SODAS software. Different approaches such as factorial analysis with supplementary elements, multiple factorial analysis and others are studied in dependence on the data. These methods do not take into account specifically the temporal aspect (time ordering) and can therefore be applied in a spatial context as well. The following time-series approaches are considered: adaptations from classical three-way data analysis, coding, data compression, alternative factor analyses, matching, and dissimilarity specifications. Marking and generalization methods are considered in the symbolic object context for developing interpretation aids in the form of symbolic objects. Several applications are described to illustrate the topic. For the entire collection see [Zbl 1039.62501].
This page was built for publication: Symbolic approaches for three-way data
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2756653)
