Imputation of Missing Values for Unsupervised Data Using the Proximity in Random Forests

摘要

This paper presents a new procedure that imputes missing values by random forests for unsupervised data. We found that it works pretty well compared with k-nearest neighbor (kNN) and rough imputations replacing the median of the variables. Moreover, this procedure can be expanded to semi-supervised data sets. The rate of the correct classification is higher than that of other conventional methods. The imputation by random forests for unsupervised or semi-supervised cases was not implemented.