Comparison of Dimensionality Reduction Methods and Strategies in Mixed-Type Data From the Healthcare Field

View at publisher Summarise with VIRAT
Authors: Emmanouil D. Pratsinakis, Zacharenia Kyrana, Nikolaos Papafilippou, Christos A. Dordas, George Menexes
Cited by: 0
References: 37

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Mixed-type data consists of a mix of both scale and categorical data within the same dataset. Analyzing mixed-type data presents unique challenges because traditional statistical methods often require variables to be of the same type. Thus, special techniques and strategies are needed to handle and analyze mixed-type data effectively. One of these strategies refers to the encoding of variables. One approach, advocated by the Dutch School of Data Analysis (GIFI System), suggests transforming categorical data into scale data to address this issue. Conversely, the French School of Data Analysis proposes converting scale data into categorical. However, these transformations may result in a loss of original data information. The aim of this study is to compare various encoding strategies to be used simultaneously in the same dimensionality reduction analysis, at the same dataset, to highlight the importance of applying different strategies of encoding data until a satisfactory solution of dimensionality reduction is found or the validity of the results is reinforced.

Fields of Study & Subjects

Statistical Methods and Applications Face and Expression Recognition Artificial Intelligence in Healthcare

Publication details

Journal / source
IGI Global eBooks
Pages
159–186
Publication date
18 March 2026
Language
NG
Document type
Book chapter