Understanding Distance Metrics in KNN Imputation: Theoretical Insights and Applications

Abstract

K-Nearest Neighbors (KNN) imputation is a widely used technique for handling missing data in machine learning and statistical modeling. The success of KNN imputation heavily depends on the choice of distance metric, as it determines the "closeness" of neighbors. This paper provides a comprehensive overview of the key distance metrics used in KNN imputation, including their theoretical background, mathematical formulations, use cases, and the implications of their selection on imputation outcomes [1-6].