I have a dataset with 13 variables and 50 observations representing the U.S. states. The variables represent the land use intensity of different agricultural industries in each state. Of those 650 values, 42 are missing. Ideally, I would like to impute the missing values taking the following three pieces of information into account:
- Relationships among the variables
- Values of neighboring states (considering only states that directly border the state with missing data)
- Population of neighboring states (i.e. weight the neighbors more highly if they are larger)
I am familiar with multiple imputation using the R package mice but I do not know whether it supports including spatial information in the way I would like to. If anyone knows of a way I can implement this imputation, in mice or some other way, it would be very helpful.
