HJ-Biplot-Pareto, a New Proposal for Data Hierarchization: Application to Multivariate Analysis of Monthly Precipitation in Ecuador (1960-2019) Complemented with STATIS-4 and Neutrosophic Z-Numbers

Abstract

This study proposes a new methodology developed by Morán-Herrera and Galindo-Villardón, which incorporates Pareto analysis into the HJ-Biplot model in order to establish a strategic hierarchical approach for selecting relevant variables in multivariate data sets. This integration allows for the objective identification of the elements that have the greatest impact on the total variability of the phenomenon studied, facilitating visual interpretation and decision-making in complex contexts. To demonstrate its application, a database with monthly precipitation records in Ecuador between 1960 and 2019 was used as a case study. Using HJ-Biplot-Pareto, it was possible to determine, for each decade, the months that accounted for 80% of the total rainfall volume, thus providing a clear representation of seasonal patterns and their temporal evolution. Complementarily, techniques such as STATIS-4 and the use of neutrosophic Z-numbers were incorporated, which allowed the integration of decadal structures into a consensus space and the modeling of the uncertainty and indeterminacy present in the data. The results show a nonlinear transition in seasonality, with significant fragmentation in the 1980s and 2000s, as well as signs of recovery in the 2010s. The incorporation of the Pareto criterion into the HJ-Biplot represents an original and versatile contribution, applicable to any multivariate set that requires the hierarchization of variables and the clear representation of internal relationships, without thematic or disciplinary restrictions.