create a lookup table correlating changes in attributes to changes in value
For each township, and each modeling group, create a table that shows the impact of a marginal change in each RHS predictor on predicted value. For RHS variables that enter non-linearly, we want to calculate the marginal impact at each quintile of the variable in question.
This table will be used to process appeals based solely on changes in characteristics that are used in modeling. We are effectively taking the new characteristic data and 'running it through the model' again, ensuring that properties are treated uniformly on appeal.