Evaluation of scoring index with different normalization and distance measure with correspondence analysis
Abstract: The purpose of this thesis is to analyze data for a scoring system and evaluate different normalization procedures and distance measures for correspondence analysis. When bootstrapping 100 samples and evaluating coordinates for the row and column profiles the results show that the representation of the coordinates for the row and column profiles are similar when looking at the normalizations methods separately. The individual positioning of the attributes and brands does not change. However, the scaling is differently presented and when looking at biplots, combining row and column profiles, a different mapping of the rows and column profiles can be seen. One has to be careful when choosing between the different normalization and distance measures. A guiding rule is to choose according to the underlying assumptions and according to the research objective. For this particular data, the relationship between the column profiles and row profiles are important hence symmetric normalization is preferred with euclidean distance.
AT THIS PAGE YOU CAN DOWNLOAD THE WHOLE ESSAY. (follow the link to the next page)