*5.3. Limitations*

Notwithstanding the substantial improvements of the statistical partitioning heuristic with respect to the calibration procedures, some extensions to the procedure itself and to the related research could still be made in the future. We now present a few limitations of the current research and propose several potential advances that could be made in these areas.


RP. In that way, we ge<sup>t</sup> even more specific partitions that should exhibit activities that are more strongly related. Furthermore, the extra managerial partitioning could be applied together with or instead of the statistical partitioning (i.e., if *selection* = 1).

4. Furthermore, the managerial partitioning criteria should not stay limited to PD, WP and RP. These are perhaps some of the most obvious and logical criteria, but there can still be others that might show even greater distinctive power for dividing a project into adequate partitions. These extra managerial partitioning criteria could be harvested from more empirical research, for example, into the drivers of project success. If those drivers could be reliably identified for a particular kind of project, they could also provide a good basis for grouping similar activities that thus show similar risks (and should therefore belong to the same partition).
