We note that outlier score function has a recursive definition. Given an incident z, in order to calculate the outlier score for Cell(z), we need to calculate the outlier scores of all parent cells of z. In a similar manner, we need to calculate outlier scores for grand-parents, great-grand-parents, etc. of cell z. Therefore, all cells that contain z will be evaluated. Theoretically, if we do this for all incident-level cells, then we will get outlier scores for all non-empty (containing at least one incident) cells for a given dataset.