Statistical Data Fusion for Advanced Use of Multi-source Datasets
Finished by March 31, 2012

To make precise forecasts in policy decision making or marketing, it is necessary, but difficult to get a single-source dataset in which every variables are observed for every subjects (or units). In this research project, the purpose is to develop statistical data fusion methods for multi-source datasets in which some variables are missing for subjects, by using covariate information and imputation of missing values.