DATA DIMENSIONALITY REDUCTION METHODS FOR ORDINAL DATA
Martin Prokop · 2011
From questionnaire survey we frequently get data, their values are expressed in ordinal (e.g. Likert) scale. The questionnaire contains usually a lot of questions, so we get multidimensional data matrix. To simplify calculations with the data it is useful to reduce dimensionality of the dataset. For ordinal data we use different or improved methods compared to quantitative data. This article includes the overview and comparison of dimensionality reduction methods (e.g. principal component analysis, factor analysis, multidimensional scaling, cluster analysis...). From these methods we get groups of similar variables (latent classes), in some cases we can create interpretation of these new variables.