CANCEIS is an edit and imputation system developed by Statistics Canada. It was designed based upon the Nearest-neighbour Imputation Methodology (NIM) and was first used to perform edit and imputation on census data. CANCEIS performs minimum change nearest-neighbour imputation and deterministic imputation. The NIM allows the simultaneous hot-deck imputation of numeric and categorical variables based on a single donor. The NIM identifies donors for the entire household, not only for the individuals. For each failed household, the NIM identifies a set of potential donors (nearest neighbours) which are as similar as possible to the failed household to be imputed. For each nearest neighbour, the smallest subsets of variables which, if imputed, allow the imputed record to pass the edits, are identified. One of those imputation actions is randomly selected by giving a better chance to those imputation actions that are simultaneously closer to the failed household and the potential donor.
|GSBPM code:||5.3 Review and validate
5.4 Edit and impute
|Keywords:||editing, nearest-neighbour donor imputation,
numeric and categorical variables