I started to read about Kriging and the implementation into R. In all the tutorials I found, a (regular) grid is used to do ordinary or universal kriging. My situation is as follows, I try to interpolate house prices, which of course do not lie on a regular grid. For all houses I have a set of covariates including features such as number of rooms and space. For a subset (approx. 20%), I have the corresponding house price. When doing Regression-kriging in estimating price ~ rooms + space, I cannot provide the data for the regular grid points. Is there a way of using unevenly spaced prediction points or how is this best tackled?
Have you read the help for
newdata: data frame or Spatial object with prediction/simulation locations; should contain attribute columns with the independent variables (if present) and (if locations is a formula) the coordinates with names as defined in ‘locations’
You create a
newdata data frame with the locations and attributes of your prediction locations and you get back predictions at those locations.