A research project completed in Winter 2022, supervised by Dr. Rahul Krishnan at the Machine Learning and Computational Healthcare Lab. Evaluated, through a causal inference perspective, the suboptimality of various imputation methods (such as doubly robust estimators and variational autoencoders) when applied to handle various mechanisms of missingness (i.e., MCAR, MAR and MNAR).
Some of the folders in this repository are cloned from the following repositories:
- NeuMiss: https://github.com/marineLM/NeuMiss
- MIRACLE: https://github.com/vanderschaarlab/MIRACLE
- BetaVAEImputation: https://github.com/gevaertlab/BetaVAEImputation.git
- HI-VAE: https://github.com/probabilistic-learning/HI-VAE.git
- IFAC-VAE-Imputation: https://github.com/ProcessMonitoringStellenboschUniversity/IFAC-VAE-Imputation.git