Identifying financial inclusion patterns in Mexico with multivariate analysis: creating dimensions by applying PCA to mixed variables

Abstract

This study addresses the complexity of financial inclusion analysis in Mexico due to the multidimensionality of the ENIF 2024 data and the number of qualitative variables contained, which compose the categories that limit the population's participation in the formal financial system. The main objective is the identification of patterns by applying multivariate analysis to properly select the set of dimensions that make up the categories under study and to extract their characteristics with greater accuracy. By applying Principal Component Analysis for mixed variables, the categorical variables of the survey are efficiently handled, considering that most multivariate techniques work with numerical variables. The findings demonstrate that this methodology is efficient for pattern detection and dimensionality reduction, which is useful for use in more in-depth research.