Professor Andriëtte Bekker

Modeling incomplete compositional datasets

Abstract

Incomplete compositional data analysis is hindered by the lack of likelihood-based methods that directly handle missing proportions on the simplex [1, 2]. This topic introduces an estimation technique that operates directly on the original data, thereby accommodating missing values while preserving the interpretability of the variables within their natural domain [3, 4]. Simulation studies demonstrate robust performance in parameter estimation and imputation across missingness mechanisms. To showcase its practicality, the proposed algorithm is applied to real datasets exhibiting distinct missingness patterns.  

Keywords: Censoring, Compositional, Clustering, Detection limit, Hidden values, Missing values, Mixture Models

References

1.Van den Boogaart, K. G., R. Tolosana-Delgado, and M. Bren. ”Concepts for handling zeroes and missing values in compositional data.” Proceedings of IAMG. Vol. 6. 20

2.Schafer, Joseph L. ”Analysis of incomplete multivariate data.” Chapman and Hall/CRC, 1997.

3.McLachlan, Geoffrey J., and Thriyambakam Krishnan. ”The EM algorithm and extensions.” John Wiley & Sons, 2008: 41-66.

4.Favaro, Stefano, Georgia Hadjicharalambous, and Igor Prunster. ”On a class of distributions on the simplex.” Journal of Statistical Planning and Inference 141.9 (2011): 2987-3004.