Project Management Journal

planned, whereas the complete project finished 29 days early. Moreover, most of the predecessors of activity 17 were completed right on time. This means that the buffers introduced in the planning (see Figure 1) were reduced as well and appeared to be oversized. In other words, the start dates of the successor activities could be advanced, as no organizational constraints (e.g., unavailability of a particular team or subcontractor until a certain date) occurred. Therefore, the project was executed even faster than what would result from shortening the activity durations alone. Because of the large fraction of variable costs in the project, the eventual cost is also reduced significantly, with a magnitude quite similar to the reduction in project duration (i.e., 16.6% compared with 18%).

Now reconsider Table 4. This table shows that it could already be seen from TP2 that the project was going to be both under budget and early, as both the CPI and the SPI(t) were consistently higher than 1. Again, it becomes clear that the progress data of TP1 were not yet reliable, as a CPI of 0.70 and SPI(t) of 0.41 incorrectly indicated that the project—when the performance-based EVM forecasting methods EAC-CPI and ESM-SPI(t) are applied—was going to be well over budget, and even more so, overdue. Furthermore, the RC and RD values could already be observed from the last column of Table 4 (i.e., the post-project forecasts of TP5). Because the RC and RD represent the actual project outcomes and, therefore, the optimal forecast values, they form the basis for evaluating the accuracy of the presented cost and time forecasting methods. More specifically, the mean absolute percentage error (MAPE) measure is used to this end. The generic MAPE formula is as follows:

MAPE 5 1 n 

n

t51

A2 Ft A

In this formula, A is the actual
(eventual) value and Ft is the forecasted
value at time instance t. In our case,
the time instances t 5 1,...,n represent
to as the real cost (RC) and real dura-
tion (RD) of the project, respectively.

Thus, the project came in more than € 40,000 (approximately, US$52,916) or 16.6% under budget and was completed 29 days or 18% earlier compared to the baseline estimates. Indeed, one can observe from Table 5 that all critical activities (i.e., activities 4, 17, 22, and 16) were completed significantly faster than planned—especially activity 4, which took only 22 days instead of 151 days, and activity 16, which even appeared to be superfluous (i.e., no adjustments to the furniture were needed). The critical activity for which the duration was least reduced is activity 17. Notice that this activity was only 19 days shorter than

Results and Discussion

The considered project was executed and exhibited the real activity outcomes presented in the two last columns of Table 5. The baseline costs and durations (i.e., the as-planned values) in the second and third columns were already shown in Table 1, but are again included here to allow for easier comparison. Furthermore, the project totals for baseline costs and durations are, of course, the BAC and PD, respectively.

On the project level, an eventual cost of €203,606 (approximately, US$269,350) and eventual duration of 132 days were reached (see the two last columns of project total in Table 5). From now on, these outcomes are referred