Missing Data Imputation using Optimal Transport

TL;DR

This paper introduces a novel data imputation method using optimal transport distances, which effectively handles various missing data scenarios and outperforms existing techniques.

Contribution

It proposes a new optimal transport-based loss function for missing data imputation and practical end-to-end learning methods that adapt to different distribution assumptions.

Findings

OT-based methods match or outperform state-of-the-art imputation techniques

Effective in MCAR, MAR, and MNAR missing data settings

Works well even with high percentages of missing data

Abstract

Missing data is a crucial issue when applying machine learning algorithms to real-world datasets. Starting from the simple assumption that two batches extracted randomly from the same dataset should share the same distribution, we leverage optimal transport distances to quantify that criterion and turn it into a loss function to impute missing data values. We propose practical methods to minimize these losses using end-to-end learning, that can exploit or not parametric assumptions on the underlying distributions of values. We evaluate our methods on datasets from the UCI repository, in MCAR, MAR and MNAR settings. These experiments show that OT-based methods match or out-perform state-of-the-art imputation methods, even for high percentages of missing values.