Main Metric Components in the Generation of Mixed Indicators: An Application of SGVD Methodology
Journal
Data Analysis and Optimization for Engineering and Computing Problems
EAI/Springer Innovations in Communication and Computing
ISSN
2522-8595
2522-8609
Date Issued
2020
Type
text::conference output::conference proceedings::conference paper
Abstract
The analysis of mixed principal components is presented by applying the generalized singular value decomposition methodology (GSVD). This multivariate analysis allows quantitative and qualitative analysis, combining principal component analysis with multiple correspondence analysis. The GSVD methodology is developed and applied to the data of the national survey of household income and expenditure in Mexico for the period 2016. The objective is to build an indicator of consumption patterns of Mexican households through a set of variables that consider sociodemographic aspects of households (qualitative) as well as variables that interest consumption items of households (quantitative). The results show that the indicator generated by mixed main components allows characterizing sets of households according to their sociodemographic characteristics and consumption patterns. This indicator allows a comprehensive evaluation of household profiles according to interest consumption items and defined sociodemographic variables. The results are presented by applying the Varimax rotation, which allows a better interpretation of the mixed main components generated. © Springer Nature Switzerland AG 2020.