Reliability of multivariate statistical analysis of tenders in public procurement

A. I. Velasquez Pizarro, M. Zarzo

The case of tenders is a big data problem of high dimensionality and binary variables. This study empirically verifies, with bidding data, the relationship between Cronbach's Alpha coefficient and dimensionality reduction criteria in order to compare two multivariate models: i) Principal Component Analysis (PCA) for the continuous case and ii) Multiple Correspondence Analysis (MCA) with categorical data. In both models, it is proved that the reliability of the latent structure will be the key factor to determine the efficiency of the bidding classifier when it is in an exploratory or confirmatory stage. Finally, the most appropriate Machine Learning (ML) techniques are selected for segmentation and to achieve greater efficiency in tender classification.

Keywords: Multiple Correspondence Analysis (MCA), Principal Component Analysis (PCA), Public Procurement, Auction, Tender.

Scheduled

Multivariate Analysis
November 8, 2023  12:00 PM
CC4: Room 2

Estudio empírico de las medianas difusas restringidas a la muestra

S. Palacio Vega, B. Sinova Fernández, M. Á. Gil Álvarez

Generación de outliers con propiedades específicas, a partir de un modelo de Análisis de Componentes Principales

F. J. Arteaga Moreno, A. González Cebrián, A. Folch Fortuny, A. J. Ferrer Riquelme