In spatial models, a neighbor relationship is usually binary. Two areal units are neighbors or they are not. This is called the W or weight matrix. This matrix may be standardized (e.g. row standardized). Has anyone developed a weight matrix that accounts for different strengths of neighborhood? I'm not talking about second order contiguity or distance neighborhoods. For an example, imagine that we had a neighborhood where some houses were divided by stone fences, others by wood fences, and still others by chain link. We wanted to count each neighbor, but quality it by the strength of the fence. Has anyone coded it in R or Python?