Marco Di Zio

Invited Session
Machine learning in the design, analysis and integration of sample surveys
Organizer: Daniela Marella (Sapienza Università di Roma)
Chair: Chiara Bocci (Università degli Studi Firenze)
Discussant: Fulvia Mecatti (Università degli Studi Milano Bicocca)
Room: T31
Floor: ground
Short summary: Machine learning methods are beginning to be used in various aspects of sample surveys such as weighting, nonresponse and data integration. In this session three main aspects are examined. First of all, the use of Bayesian Networks to deal with the statistical matching problem in a multivariate context is discussed. Next, machine learning techniques incorporating the sampling weights in the imputation process are considered. Finally, a modified version of the PC algorithm for structural learning is proposed to take into account complex sample designs.

Paper
Mass imputation through Machine Learning techniques in presence of multi-source data Author(s)  Fabrizio De Fausti   Marco Di Zio   Romina Filippini   Simona Toti