Data Imputation with an Autoencoder and MAGIC

• DOI: 10.1109/sampta59647.2023.10301413
• Publication Date: 2023-11-02T17:47:46Z
• Authors: Devin Eddington; Andres F. Duque C...; Guy Wolf; Kevin R. Moon
• Keywords: Imputation (statistics); Autoencoder; Robustness

Missing data is a common problem in many applications. Imputing missing values challenging task, as the imputations need to be accurate and robust avoid introducing bias downstream analysis. In this paper, we propose an ensemble method that combines strengths of manifold learning-based imputation called MAGIC autoencoder deep learning model. We call our Deep MAGIC. trained on linear combination mean squared error original MAGIC-imputed data. Experimental results three benchmark datasets show outperforms several state-of-the-art methods, demonstrating its effectiveness robustness handling large amounts