Iterative Imputation of Missing Data Using Auto-Encoder Dynamics

Detracking Autoencoding Conditional Generative Adversarial Network: Improved Generative Adversarial Network Method for Tabular Missing Value Imputation

“…In addressing missing values within datasets, the mainstream approaches fall into three main categories: no treatment, direct deletion, and imputation methods. For datasets with a small proportion of missing values, direct mining and analysis of the original data containing null values can be performed, utilizing methods such as Bayesian networks [6][7][8]. The direct deletion method involves removing data objects, attributes, or paired variables that contain missing values [9].…”

Estimating conditional density of missing values using deep Gaussian mixture model

“…The authors of [15,23] define a sampling procedure based on pseudo-Gibbs sampling and Metropolis-within-Gibbs algorithm for filling missing values by iterative autoencoding of incomplete data. Śmieja et al [25,26] propose iterative algorithm for A preliminary version of this paper appeared as an extended abstract [21] at the ICML Workshop on The Art of Learning with Missing Values. maximizing conditional density based on the dynamics of auto-encoder models.…”

How to ask twenty questions and win: Machine learning tools for assessing preferences from small samples of willingness-to-pay prices

“…Principled rules for data cleaning can help remove these outlying response times (Rahm & Do, 2000), but then we are left with the issue of determining what to do with the missing values. While there are methods to impute missing data or handle missing values, such as by using common values in the second layer of the network (Sharpe & Solly, 1995;Śmieja et al, 2018), it is ideal to not have to deal with outliers in the first place.…”