%0 Book Section %1 WOLFF199885 %A Wolff, Karl Erich %A Gabler, Siegfried %B Visualization of Categorical Data %C San Diego %D 1998 %E Blasius, Jörg %E Greenacre, Michael %I Academic Press %K distance fca psychology scaling %P 85-97 %R https://doi.org/10.1016/B978-012299045-8/50008-5 %T Chapter 7 - Comparison of Visualizations in Formal Concept Analysis and Correspondence Analysis %U https://www.sciencedirect.com/science/article/pii/B9780122990458500085 %X Publisher Summary This chapter compares visualizations in the formal concept analysis (FCA) and correspondence analysis (CA) in its two variants: simple correspondence analysis (CA) and multiple correspondence analyses (MCA). In contrast to CA, the data evaluation in FCA starts with a many-valued context given by the original data and represents an ordinal structure on the values of each many-valued attribute by the scaling procedure. The graphic representation of the line diagrams of the concept lattice contains all information about the derived context. Hence FCA has an exact graphic data representation in contrast to the metric approximation of the data in CA. However the exact representation has the disadvantage that even small many-valued contexts may have concept lattices with thousands of concepts. An alternative technique is to use nested line diagrams that can be automatically generated from the small diagrams of the scales. It is suggested to use FCA for data with a small number of many-valued attributes. For data with more than 20 many-valued attributes, MCA should be applied to the suitably scaled context to find some interesting attribute clusters that may serve as the starting point for a data analysis with FCA. %@ 978-0-12-299045-8

@incollection{WOLFF199885, abstract = {Publisher Summary This chapter compares visualizations in the formal concept analysis (FCA) and correspondence analysis (CA) in its two variants: simple correspondence analysis (CA) and multiple correspondence analyses (MCA). In contrast to CA, the data evaluation in FCA starts with a many-valued context given by the original data and represents an ordinal structure on the values of each many-valued attribute by the scaling procedure. The graphic representation of the line diagrams of the concept lattice contains all information about the derived context. Hence FCA has an exact graphic data representation in contrast to the metric approximation of the data in CA. However the exact representation has the disadvantage that even small many-valued contexts may have concept lattices with thousands of concepts. An alternative technique is to use nested line diagrams that can be automatically generated from the small diagrams of the scales. It is suggested to use FCA for data with a small number of many-valued attributes. For data with more than 20 many-valued attributes, MCA should be applied to the suitably scaled context to find some interesting attribute clusters that may serve as the starting point for a data analysis with FCA.}, added-at = {2022-02-09T17:14:46.000+0100}, address = {San Diego}, author = {Wolff, Karl Erich and Gabler, Siegfried}, biburl = {https://www.bibsonomy.org/bibtex/2b59ac40917f951ea2e8ee3615da3711a/tomhanika}, booktitle = {Visualization of Categorical Data}, doi = {https://doi.org/10.1016/B978-012299045-8/50008-5}, editor = {Blasius, Jörg and Greenacre, Michael}, interhash = {331cfae395d2d3b84e0534d261e16199}, intrahash = {b59ac40917f951ea2e8ee3615da3711a}, isbn = {978-0-12-299045-8}, keywords = {distance fca psychology scaling}, pages = {85-97}, publisher = {Academic Press}, timestamp = {2022-02-09T17:14:46.000+0100}, title = {Chapter 7 - Comparison of Visualizations in Formal Concept Analysis and Correspondence Analysis}, url = {https://www.sciencedirect.com/science/article/pii/B9780122990458500085}, year = 1998 }