Imputation techniques for incomplete data in quadratic discriminant analysis

Abstract

We have compared the efficacy of five imputation algorithms readily available in SAS for the quadratic discriminant function. Here, we have generated several different parametric-configuration training data with missing data, including monotone missing-at-random observations, and used a Monte Carlo simulation to examine the expected probabilities of misclassification for the two-class quadratic statistical discrimination problem under five different imputation methods. Specifically, we have compared the efficacy of the complete observation-only method and the mean substitution, regression, predictive mean matching, propensity score, and Markov Chain Monte Carlo (MCMC) imputation methods. We found that the MCMC and propensity score multiple imputation approaches are, in general, superior to the other imputation methods for the configurations and training-sample sizes we considered.

Alternativas de análise para experimentos G × E multiatributo

“…Por outro lado, tipicamente um pequeno número de imputações (3 ≤ M ≤ 10)é necessário para obter um bom desempenho da IM (Ounpraseuth et al, 2012) e se utilizará M = 5, uma vez que esse número permite atingir alta eficiência estatística em muitas aplicações práticas (van Buuren, 2012).…”