Multiple imputation with compatibility for high-dimensional data

Multiple Imputation (MI) is always challenging in high dimensional settings. The imputation model with some selected number of predictors can be incompatible the analysis leading to inconsistent and biased estimates. Although compatibility such cases may not achieved, but one obtain consistent unbiased estimates using a semi-compatible model. We propose relax lasso penalty for selecting large set variables (at most n). substantive that also uses formal variable selection procedure high-dimensional structures then expected nested this resulting will probability. likelihood unstable face convergence issues as becomes nearly sample size. To address these issues, we further use ridge obtaining posterior distribution parameters based on observed data. proposed technique compared standard MI software techniques available data simulation studies real life dataset. Our results exhibit superiority approach existing approaches while addressing issue.