The assessment of project complexity is important in order to predict the outcome of a project scheduling and/or its monitoring. The evaluation of this complexity seems difficult especially with the presence of working time flexibilities as well the multi-skilled workforce with heterogeneous productivity level that differs from skill to another. In this paper, we present a group of different measures that can be used to quantify the numerous characteristics of a project and the required resources. Therefore, the most significant quantifiers of the network regarding to its size, dependencies, shape, asymmetry, and its bottlenecks will be presented. Moreover, the quantifiers related to the temporal characteristics, work-content, and availability of resources will also be discussed and presented. Relying on the normalised version of these measures and using a data set of projects with different descriptions of work-contents and resources availabilities, the smallest number of project complexity indices will be produced. Linear aggregations of these indices were conducted using the principal components analysis. Subsequently, these indices were used in evaluating the performance and robustness of a metaheuristics approach that used genetic algorithms. The result analysis showed the effectiveness of some of the proposed indices to explain the most variance of different outcomes from a data set of 400 projects. Moreover, one of these indices that we called “project weight index”, can be used efficiently to predict the presence of lateness penalties or the un-capability to realise the project with the specified resources during a given specified duration. |
