scientific article; zbMATH DE number 1368019
This paper deals with the modern state and new trends of factor analysis of qualitative data in presence of a dependence structure and lack of distribution hypotheses. The main task is to present an overview of the contribution of Italian researchers in developing original methodology named ``non-symmetrical correspondence analysis (NSCA) which gives the possibility to include a priori information in multidimensional data analysis techniques, giving special attention to categorical variables. NSCA methodology, based on decomposition of a predictability index \(\tau _b\), enables the analysis of contingency tables when the behaviour of one variable is supposed to be dependent on the other cross-classified variable. The outline of this method and its main properties are given as well as extensions for analysis of multiple tables. So, multiple and partial versions of NSCA, based on multiple or partial \(\tau _b\)'s, are proposed and the forward selection procedure for choosing variables with higher predictive power is presented. All theoretical statements are illustrated by numerical examples.
- Some interpretative tools for non-symmetrical correspondence analysis
- Regularized nonsymmetric correspondence analysis
- Tests of ignoring and eliminating in nonsymmetric correspondence analysis
- Visualizing main effects and interaction in multiple non-symmetric correspondence analysis
- Analysing dependence in large contingency tables: Nonsymmetric correspondence analysis and regression with optimal scaling
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4702172)