**4. Partitioning Heuristic**

The improved acceptance rate of the extended calibration method as well as its limitations have been the main driving force to develop the new statistical partitioning heuristic. It integrates the hypothesis testing approach of the original calibration method with the human partitioning philosophy of the extended calibration method, and, consequently, follows a similar methodology as both calibration procedures. The main difference is that the statistical partitioning method now partitions the project data not only based on human input, but also using a statistical methodology, and this extension has resulted in a number of significant modifications graphically summarized in Figure 4.

In this section, an overview of the newly developed partitioning method will be given subdivided in three main subsections. Each of these three sections overcome (some of) the limitations that still existed for the extended calibration procedure. We will not run through the solution approach in an explicit stepwise manner as was done in Section 3.1, but rather show where the steps (S0 to S4) have been incorporated and possibly adapted. In Section 4.3.3, the discussed limitations will be addressed chronologically and referred to when and how a particular option of the statistical partitioning heuristic solves them.
