Missing Data Imputation in Multivariate Analysis

  • Mario Romanazzi
Conference paper
Part of the Studies in Classification, Data Analysis, and Knowledge Organization book series (STUDIES CLASS)


A new imputation method for incomplete data is suggested, to be used with non-random multivariate observations. The principle is to fill the gaps in a data set so that the partial complete-data configuration and the total filled-in configuration are similar, according to a matrix correlation coefficient. The optimality criteria are Escoufier’s RV and Procrustes normalized statistic. Three examples are illustrated.


Imputation Method Artificial Data Iterative Optimization Ordinary Linear Regression Miss Data Imputation 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. Crettaz De Roten F. and Helbling, J.-M. (1991): Une estimation de données manquantes basée sur le coefficient RV. Revue de Statistique Appliquée, 39, 47–57.Google Scholar
  2. Everitt, B. S. (1984): An. introduction to latent variable models. Chapmann and Hall. Romanazzi, M. (1995): Missing values imputation and matrix correlation. Quaderni di Statistica e Matematica applicata alle Scienze Economico-Sociali, 15, 41–59.Google Scholar
  3. Seber, G. A. F. (1984): Multivariate Observations. Wiley.Google Scholar

Copyright information

© Springer Japan 1998

Authors and Affiliations

  • Mario Romanazzi
    • 1
  1. 1.Department of Statistics“Ca’ Foscari” University of VeniceVeniceItaly

Personalised recommendations