Handling datasets with partially observed variables poses substantial methodological challenges, especially when the available information is scarce or nearly absent. Traditional imputation techniques may introduce bias, as reliance on the observed data can lead to an overrepresentation of less frequent subpopulations. In this paper, we propose a mixture-model–based approach to identify and characterize key differences between partially observed and completely unobserved subpopulations. The effectiveness of the proposed method is demonstrated through a simulation study and an educational case study. .
Nicolussi, F., Masci, C., Bertarelli, G., Terzera, L., Mecatti, F. (2026). Mixture Models for Partially Observed Variables. In F. Martella, S. Arima, M.F. Marino, C. Mollica (a cura di), Statistical Science: From Theory to Applied Research IV SIS-FENStatS 2026, Short Papers, Contributed Sessions 3 (pp. 118-123). Springer [10.1007/978-3-032-30665-4_20].
Mixture Models for Partially Observed Variables
Terzera,L;Mecatti, F
2026
Abstract
Handling datasets with partially observed variables poses substantial methodological challenges, especially when the available information is scarce or nearly absent. Traditional imputation techniques may introduce bias, as reliance on the observed data can lead to an overrepresentation of less frequent subpopulations. In this paper, we propose a mixture-model–based approach to identify and characterize key differences between partially observed and completely unobserved subpopulations. The effectiveness of the proposed method is demonstrated through a simulation study and an educational case study. .
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
