Mattia Spaziani (e-mail: spaziani@istat.it)
Italian National Institute of Statistics (Istat), Rome, Italy
Doriana Frattarola (e-mail: frattarola@istat.it)
Italian National Institute of Statistics (Istat), Rome, Italy
Marcello D’Orazio (e-mail: madorazi@istat.it)
Italian National Institute of Statistics (Istat), Rome, Italy
Food and Agriculture Organization of the United Nations, Rome, Italy
Abstract
This work introduces a relatively new procedure for integrating social survey data through combined use of machine learning techniques and well-known statistical matching methods. The integration is performed with the scope of studying the relationship between variables not jointly observed in the same survey. The results of the new matching procedure seem promising since they are better than those provided by traditional matching methods.
Keywords: statistical matching, household surveys, supervised classification
JEL classification: C40, C80, I31
